Nitrogen isotopes link mycorrhizal fungi and plants to nitrogen dynamics

Authors


Author for correspondence:

Erik A. Hobbie

Tel: +1 603 862 3581

Email: erik.hobbie@unh.edu

Abstract

Contents

 Summary367
I.Introduction367
II.Background on isotopes368
III.Patterns of soil δ15N370
IV.Patterns of fungal δ15N372
V.Biochemical basis for the influence of fungi on δ15N patterns in plant–soil systems373
VI.Patterns of δ15N in plant and fungal culture studies374
VII.Mycoheterotrophic and parasitic plants375
VIII.Patterns of foliar δ15N in autotrophic plants376
IX.Controls over plant δ15N377
X.Conclusions and research needs378
 Acknowledgements379
 References379

Summary

In this review, we synthesize field and culture studies of the 15N/14N (expressed as δ15N) of autotrophic plants, mycoheterotrophic plants, parasitic plants, soil, and mycorrhizal fungi to assess the major controls of isotopic patterns. One major control for plants and fungi is the partitioning of nitrogen (N) into either 15N-depleted chitin, ammonia, or transfer compounds or 15N-enriched proteinaceous N. For example, parasitic plants and autotrophic hosts are similar in δ15N (with no partitioning between chitin and protein), mycoheterotrophic plants are higher in δ15N than their fungal hosts, presumably with preferential assimilation of fungal protein, and autotrophic, mycorrhizal plants are lower in 15N than their fungal symbionts, with saprotrophic fungi intermediate, because mycorrhizal fungi transfer 15N-depleted ammonia or amino acids to plants. Similarly, nodules of N2-fixing bacteria transferring ammonia are often higher in δ15N than their plant hosts. N losses via denitrification greatly influence bulk soil δ15N, whereas δ15N patterns within soil profiles are influenced both by vertical patterns of N losses and by N transfers within the soil–plant system. Climate correlates poorly with soil δ15N; climate may primarily influence δ15N patterns in soils and plants by determining the primary loss mechanisms and which types of mycorrhizal fungi and associated vegetation dominate across climatic gradients.

I. Introduction

Nitrogen (N) commonly limits plant growth in terrestrial ecosystems (Tamm, 1991; Vitousek & Howarth, 1991) and how plants obtain N influences both vegetational responses to elevated CO2 (Finzi et al., 2007) and competitive interactions within plant communities (Wilson & Tilman, 1991). Most plants rely on symbiotic (mycorrhizal) fungi to supply them with N. These fungi function at the interface between plants and the soil from which plants derive their nutrients, with much of the carbon (C) flux from plants to the soil mediated by mycorrhizal fungi. Because of the increased surface area for nutrient absorption of fungal hyphae and the extensive enzymatic capabilities of many mycorrhizal fungi, the dominant plants in most ecosystems are mycorrhizal.

Mycorrhizal fungi can be separated into three major groups, arbuscular mycorrhizal, ectomycorrhizal, and ericoid mycorrhizal fungi. These fungal types differ considerably in their spatial extent, C demands, type of host plant, species diversity, enzymatic capabilities to access different forms of N, and association with different N dynamics (Read & Perez-Moreno, 2003; Smith & Read, 2008; He et al., 2009). In general, arbuscular mycorrhizal fungi are adapted to sites with high rates of N cycling and lack proteolytic capabilities, ericoid mycorrhizal fungi are adapted to organic-rich ecosystems with low rates of N cycling and have good proteolytic and litter-degrading capabilities, and ectomycorrhizal fungi vary greatly in their enzymatic capabilities. Ectomycorrhizal taxa with hydrophilic ectomycorrhizas often lack proteolytic capabilities and assimilate soluble N forms after release from free-living microbes, whereas other ectomycorrhizal taxa with hydrophobic ectomycorrhizas have proteolytic capabilities and are adapted to N-limited conditions (Lilleskov et al., 2011). Rates of N decomposition may therefore determine the extent to which plants rely on N mobilized by mycorrhizal enzymes for N supply. Many saprotrophic fungi and many ectomycorrhizal fungi with hydrophobic ectomycorrhizas also form aggregated hyphae (rhizomorphs) for long-distance transport, which is presumably an adaptation for patchily distributed resources (Peay et al., 2011).

One common technique for assessing N dynamics, measuring N isotope ratios (15N/14N) in plants and other ecosystem pools, has proved useful because it reflects the sum total of inputs, outputs, and fractionation processes. In addition, interpretations of N isotope ratios at natural abundance do not need to consider experimental artifacts, which may be a problem in N isotope tracer studies. Despite these advantages, the interpretation of N isotope patterns in nature remains an inexact science because of the range of fluxes potentially influencing specific pools, difficulties in studying the soil processes that control many key N transformations, and the uncertainty of isotopic fractionation during many of these processes (Robinson, 2001). As a result, although quantitative and analytical relationships among δ15N of different ecosystem components have been proposed, scientists have lacked underlying theories to explain N isotope patterns that are as widely accepted as those used to interpret plant isotope patterns of C (Farquhar et al., 1982), hydrogen, and oxygen (Roden et al., 2000).

Mycorrhizal fungi appear to be key mediators of N movement in the plant–soil system that can influence isotopic patterns. Mycorrhizal fungi provide plants with access to organic N forms that are usually higher in 15N than the inorganic forms generally considered available to plants. Mycorrhizal fungi also retain 15N-enriched N and transfer 15N-depleted N to plant hosts. They also influence N availability by competing against saprotrophic (free-living) fungi and bacteria for N and assimilating ammonium and organic N before it can be mineralized or nitrified. Mycorrhizal fungi therefore influence ecosystem N loss rates via denitrification and nitrification, processes that discriminate strongly against 15N. We accordingly propose that interactions between plants and mycorrhizal fungi greatly influence both plant and soil δ15N. Many ectomycorrhizal fungi (but not other mycorrhizal fungi) produce macroscopic reproductive structures (sporocarps) that can be readily sampled for isotopic analyses, providing additional insights into their functioning.

A previous Tansley review by Högberg (1997) laid out both general principles and numerous case studies on N isotope patterns from individual systems. Since then, several broad surveys have been published of δ15N patterns in foliage, soil, and fungi (Handley et al., 1999; Amundson et al., 2003; Craine et al., 2009; Hobbie & Ouimette, 2009; Mayor et al., 2009). In addition, several reviews have addressed different aspects of N isotope interpretation, with general overviews by Evans (2001), Robinson (2001), and Adams & Grierson (2001), the biochemical basis for N isotope patterns in Werner & Schmidt (2002), the controls over fungal δ15N by Hobbie (2005), and using natural abundance and tracer 15N to assess transfers of N in fungal–plant networks by He et al. (2009).

In this review, we combine insights from culture studies and site-specific field studies with data from broad surveys to assess whether mycorrhizal fungi drive N isotope patterns in terrestrial plants and soils. We also explore the links between functional characteristics and δ15N in different taxa of ectomycorrhizal fungi. The δ15N patterns in mycoheterotrophic plants (reliant on fungi for their C and nutrients), parasitic plants (tapping into the xylem or phloem of other plants), and N2-fixing plants are also examined for potential additional insights into mechanisms governing isotopic patterns in symbioses.

II. Background on isotopes

Nitrogen has two stable isotopes, 15N (0.3663% of total) and 14N (99.6337% of total). Most physical, chemical and biochemical processes favor the initial incorporation of the lighter isotope in the product, leaving the substrate enriched in the heavy isotope. The magnitude of this ‘isotopic fractionation’ differs depending on the compounds involved and the specific reaction mechanism. Isotopically fractionating processes therefore result in differences in the isotopic ratios between the substrate and the product. These ratios depend on the isotopic ratio of the substrate, the proportion of substrate transformed to product, and whether the system is open or semi-closed (Fig. 1). Isotope effects during reactions can be expressed mathematically in several ways. In one convention, isotopic fractionation is expressed as the rate reaction or equilibrium constant of the light isotope divided by that for the heavy isotope, such that α = 14K/15K. The reverse convention of α = 15K/14K is occasionally reported.

Figure 1.

Nitrogen (N) isotope patterns in an open vs a semi-closed system. In this example, the substrate, available N, is taken up by fungi, then a fraction (transfer ratio, Tr) is transferred to host plants (product pool) with a fractionation of Δf during creation of transfer compounds (a). At a value for available N of 0‰ and for Δf of 10‰, this leads to the isotopic patterns shown in open systems (b), and semi-closed systems (c), as Tr is varied from 0 to 1. The remaining fraction (1 − Tr) is used to construct fungal biomass. Modified from Hobbie & Agerer (2010). With kind permission from Springer Science and Business Media.

Natural abundance studies use isotopic differences among different ecosystem pools and compounds to understand the sources and fluxes of many biologically important compounds. Because differences in isotopic ratios are very small, they are measured using the ‘δ’ notation, as deviations in parts per mil (‰) from a standard ratio, according to Eqn (Eqn 1).

display math(Eqn 1)

In Eqn (Eqn 1), R equals the molar abundance of the heavy isotope divided by the light isotope (15N/14N). The isotopic standard for N is atmospheric N2 (15N/14N = 0.0036765; Mariotti, 1983; Hoefs, 1997). For δ15N, pools in terrestrial ecosystems are usually between 20 and −10‰. In comparisons among samples, samples with more of the heavy isotope are commonly referred to as isotopically enriched, or heavy, and samples with less of the heavy isotope are referred to as isotopically depleted, or light. Isotopic fractionation in per mil is expressed as Δ, with

display math(Eqn 2)

If a steady state is assumed, isotopic fractionation (Δ) for reactions in open systems can be calculated based on the isotopic signature of the source and product according to Eqn (Eqn 3).

display math(Eqn 3)

An important distinction in interpreting isotopic patterns is whether kinetic or equilibrium isotopic effects dominate in a reaction. In kinetic isotopic effects, the light isotope reacts more rapidly than the heavy isotope, so that the resulting product is depleted in the heavy isotope relative to the substrate. Isotopic effects during irreversible reactions are governed by kinetic isotopic effects. By contrast, in equilibrium reactions the back-reaction from product to substrate also occurs, and therefore the kinetic isotope effects for both the forward and back reactions must be considered. In equilibrium reactions the heavy isotope generally concentrates in the compound with stronger bonds (Bigeleisen, 1965). For example, in the equilibrium reaction between ammonia and ammonium, the ammonium ion accumulates 19–21‰ more 15N than ammonia, whereas the kinetic isotope effect of the forward reaction is c. 27‰. Similarly, the equilibrium isotope effect in a reaction is the sum of the forward and the backward kinetic isotopic effects; for example, for the reaction glutamate + oxaloacetate → alpha-oxoglutarate + aspartate as catalyzed by oxaloacetate:glutamate aminotransferase, the forward reaction has a kinetic isotope effect (designated α) of 1.0083 and the back reaction a kinetic effect of 1.0017, so that the equilibrium isotope effect is 1.0083/1.0017 = 1.0066, or a 6.6‰ discrimination against 15N (Macko et al., 1986).

Information about patterns of 15N discrimination in biochemical reactions can be useful. However, translating such information into insights into the behavior of bulk samples is not easy. Both kinetic and equilibrium isotopic effects are often reported, with equilibrium effects equivalent to adding the kinetic effects for the forward and the back reactions of an equilibrium reaction. Some of the most important reactions for isotopic distributions are presented in Table 1. Equilibrium isotope effects were summarized in Rishavy & Cleland (1999) and kinetic isotope effects were summarized in Werner & Schmidt (2002). An additional factor to consider when applying isotopic discrimination factors is whether to treat a system as open or closed, as for the same isotopic discrimination factor the resulting distribution of isotopic values will differ. Metabolic branch points, where different metabolic products are produced from a common precursor substrate, are also required for isotopic differences among different pools, as pointed out for C isotope distributions by Hayes (2001).

Table 1. 15N fractionation during biochemical reactions or 15N enrichment between protein and chitin
ReactionValue (‰)SourceReferences
  1. Here, laboratory studies were all of kinetic isotope effects, whereas field studies were of net measured fractionations.

  2. a

    Chymotrypsin.

  3. b

    Papain.

  4. c

    Glutamate → aspartate.

  5. d

    Aspartate → glutamate.

Nitrogen fixation0–2Högberg (1997); Clarkson et al. (2005)
Denitrification13.2FieldHoulton et al. (2006)
Denitrification13–30FieldPérez et al. (2000)
Mineralization2.4FieldKoba et al. (2003)
Nitrification11.4FieldKoba et al. (2003)
Protein – chitin9.9FieldTaylor et al. (1997)
Protein – chitin8–10FieldHobbie & Colpaert (2003)
Protein – chitin7, 15FieldHobbie et al. (2012)
Polypeptide cleavage10a, 24bLaboratoryMedina & Schmidt (1982)
Urea hydrolysis8LaboratoryWerner & Schmidt (2002)
Arginine hydrolysis10LaboratoryWerner & Schmidt (2002)
Ammonium assimilation9–20LaboratoryHandley & Raven (1992)
Transamination8.3c, 1.7dLaboratoryWerner & Schmidt (2002)

Different mathematical models have been created to explain isotopic partitioning in two-component systems that are considered to be open (exogenous inputs to the system) or closed (no inputs to the system; Hayes, 2001). As an example of an open system (Fig. 1b), the proportion of N taken up by mycorrhizal fungi that is then passed on to host plants, termed the transfer ratio (Tr), can explain N isotope patterns in mycorrhizal plants according to the following equation (Hobbie et al., 2000):

display math(Eqn 4)

with the transfer ratio, Tr, between 0 and 1, and the discrimination against 15N during the creation of transfer compounds (Δ) of 8–10‰. To explore isotope patterns among plants, available N, and mycorrhizal fungi in an open system, Eqn (Eqn 3) can be rearranged, with δ15Nplant = δ15Nproduct, and δ15Navailable nitrogen = δ15Nsubstrate. As a simplification, the denominator (1 + δ15Nplant) can be dropped, as with a value of δ15Nplant usually between –10 and 10‰, the denominator is generally between 0.99 and 1.01. We previously used N budgets and 15N differences among available N, nonmycorrhizal pines and mycorrhizal pines under controlled conditions (Hobbie & Colpaert, 2003) to estimate Δ at 9‰. The 15N content of mycorrhizal fungi can then be calculated according to Eqn (Eqn 5).

display math(Eqn 5)

In an intermediate case, the semi-closed system, the fungal pool is not replenished by additional N from the available N pool. In this system (Fig. 1c), Eqns (Eqn 6) and (Eqn 7) can be used to calculate the isotopic signatures of plants and fungi, respectively (modified from Hobbie et al., 2005). The quantity f is the proportion of original substrate transformed to product, and is equivalent to the transfer ratio (Tr) in Eqns (Eqn 4) and (Eqn 5).

display math(Eqn 6)
display math(Eqn 7)

This approach can be used directly on data from laboratory settings, where δ15Navailable nitrogen is known. In field settings, the multiple N sources have dynamically varying isotopic signatures, which restricts the direct application of these equations.

III. Patterns of soil δ15N

1. Bulk soil

Bulk soil δ15N in a steady state must reflect the input and output δ15N signatures (Amundson et al., 2003). Thus, if mycorrhizal type or soil age correlates with soil δ15N, then this presumably reflects parallel correlations between these parameters and the proportion of N lost via highly fractionating reactions, such as denitrification and nitrification. Houlton et al. (2009) showed that streamwater nitrate was similar to bulk soil, and denitrification losses of N2 and N2O must therefore be the dominant influence on system 15N balances.

Large-scale databases for soil δ15N were compiled by Martinelli et al. (1999), Handley et al. (1999) and Amundson et al. (2003). Bulk soil δ15N averages between 2 and 6‰ in most cases, with soil δ15N values commonly increasing with greater depth in the soil profile. Amundson et al. (2003) integrated δ15N values for the top 50 cm of soil, and then regressed these values against site temperature and precipitation for both their entire data set and a subset of their data set (termed the climosequence data) for which mountain sites were selected specifically along gradients of precipitation and temperature. This climosequence regression (adjusted r2 = 0.36; = 30) was then applied to global patterns of precipitation and temperature to create a global map of estimated soil δ15N. Subsequent work has applied this regression to calculate global patterns of trace gas fluxes from soil, under the assumption that soil δ15N is at steady state, and reflects the fractionation against 15N of the major loss terms, and that the calculated regression is a reasonable representation of the influence of precipitation and climate on patterns of N losses and global patterns of soil δ15N (Houlton et al., 2009).

Soil δ15N was poorly correlated with climate for the complete data set (adjusted r2 = 0.16; = 48), and very poorly correlated with the data that were not part of the climosequence data set (adjusted r2 = 0.05; = 18). Thus, the regression model of Amundson et al. (2003) predicted soil δ15N reasonably well for the specified mountain locations, but lacked predictive power outside of these sites. Accordingly, using this regression to estimate global patterns of δ15N and trace gas fluxes (denitrification in Houlton et al., 2009) is not valid in our view. We conclude that factors other than climate must influence the long-term fluxes of 15N-depleted N from soil.

Because mycorrhizal type varies strongly with climate (Table 2) and is correlated with patterns of N cycling (Read & Perez-Moreno, 2003), the hidden influence of mycorrhizal association is a potentially confounding factor. We reanalyzed the climosequence data of Amundson et al. (2003) by adding mycorrhizal type as an additional variable, with sites classified as dominated by nonmycorrhizal, ectomycorrhizal, or arbuscular mycorrhizal plants, and a few savannah sites classified as mixed arbuscular mycorrhizal/ectomycorrhizal. The regression analyses on the Amundson et al. (2003) data are presented in Table 3. Adding mycorrhizal type to the model increased the strength of the regression, with the adjusted r2 increasing from 0.35 to 0.52 for climosequence data and from 0.16 to 0.27 for the entire data set. Temperature explained the smallest amount of variance. Thus, precipitation and mycorrhizal type are more important than temperature in influencing the δ15N of bulk soil over climatic gradients.

Table 2. Mean annual temperature (MAT; °C) and precipitation (MAP; mm) ± SE for sites where foliage of different mycorrhizal types was collected in the Craine et al. (2009) database of foliar δ15N
Mycorrhizal typeMATMAP n
  1. n is the number of each mycorrhizal type in the database.

Nonmycorrhizal1.3 ± 0.5439 ± 14702
Arbuscular mycorrhizal18.5 ± 0.11162 ± 98609
Ectomycorrhizal4.5 ± 0.2983 ± l31824
Ericoid mycorrhizal0.7 ± 0.4604 ± 19692
Table 3. Mycorrhizal type and climate correlate with the δ15N of bulk soil (0–50 cm)
ParameterClimosequenceClimosequenceAll SamplesAll Samples
 No mycorrhizalMycorrhizalNo mycorrhizalMycorrhizal
  1. Climate and bulk soil data are from Amundson et al. (2003); mycorrhizal type is based on dominant vegetation, and classified as arbuscular mycorrhizal (AM) (32), ectomycorrhizal (ECM) (11), mixed arbuscular/ectomycorrhizal (AM/ECM) (4), and nonmycorrhizal (1) (Supporting Information Table S1). Regressions include either the complete data set (= 48) or just climosequence data (= 30) from mountain gradients of temperature or precipitation. Regressions were done with or without mycorrhizal type as an explanatory variable. Mycorrhizal type significantly correlated with bulk δ15N for the climosequence (P = 0.009) and the complete data set (P = 0.033). Significance values indicated in parentheses after coefficients (± SE) for different variables. e−3 = 10−3. Regressions were done using jmp 9.0 (SAS, Cary, NC, USA). MAT, mean annual temperature (°C); MAP, precipitation (mm).

Adjusted r20.360 (< 0.001)0.527 (< 0.001)0.159 (0.008)0.267 (0.002)
Intercept4.4 ± 1.14.9 ± 1.33.9 ± 1.12.8 ± 1.3
MAT0.17 ± 0.06 (0.010)0.15 ± 0.09 (0.136)0.20 ± 0.07 (0.006)0.14 ± 0.07 (0.057)
MAP−0.7 ± 0.4e−3 (0.002)−1.6 ± 0.3e−3 (< 0.001)−1.2 ± 0.3e−3 (0.073)−1.01 ± 0.38e−3(0.010)
AM1.67 ± 0.64 (0.015)2.65 ± 0.87 (0.004)
AM/ECM−2.12 ± 0.73 (0.007)
ECM−0.22 ± 1.20 (0.858)1.54 ± 1.05 (0.151)

Age may be an additional factor influencing bulk soil δ15N. For example, tropical ultisols and inceptisols in Costa Rica and Panama (apparently dominated by arbuscular mycorrhizal trees) were depleted by 4–5‰ in 15N relative to old and highly weathered Brazilian oxisols (Pérez et al., 2000; Corre et al., 2010). In general, older soils in chronosequence studies have higher δ15N values (Brenner et al., 2001; Menge et al., 2011), presumably as a consequence of N limitation decreasing and losses of nitrate, N2O, and N2 increasing as phosphorus availability decreases.

The effects of denitrification and nitrification on soil 15N enrichment were explored by Houlton et al. (2006, 2007) in Hawaii in two detailed studies. In these studies, N isotope patterns in plants, soil, and inorganic N across a precipitation gradient from 2200 to 5000 mm indicate that sites with higher rainfall may allow essentially all produced nitrate to be denitrified, leading to no 15N enrichment of soils. By contrast, with partial denitrification at lower rainfall, 15N-enriched residual substrate was reassimilated, leading to 15N enrichment of the resulting soils.

2. Soil profiles

Bulk δ15N generally increases within soil profiles and provides information about what processes control N movement and retention. This increase varies with the dominant mycorrhizal type; δ15N increased with depth an average of 9‰ in ectomycorrhizal systems and 4‰ in arbuscular mycorrhizal systems (measured from the litter to 50 cm depth; Hobbie & Ouimette, 2009). In 40% of arbuscular mycorrhizal systems, soil δ15N increases initially and then declines at greater depths. This presumably reflects the importance in some arbuscular mycorrhizal systems of 15N-depleted products, such as dissolved organic N or nitrate, which can be reassimilated at greater depths. In addition, the loss of 15N-depleted N2O and N2 may enrich the remaining N in 15N primarily at depths favorable for denitrification. When mycorrhizal type is ignored, mean annual temperature (MAT) and precipitation (MAP) weakly correlate (adjusted r2 = 0.17) with the difference in δ15N between soil at 50 cm depth and litter (Hobbie & Ouimette, 2009) according to the following equation, with MAP measured in millimeters:

display math(Eqn 8)

However, when soils are classified by mycorrhizal type into arbuscular mycorrhizal (AM), ectomycorrhizal (ECM), and mixed arbuscular/ectomycorrhizal systems, temperature and precipitation no longer influence the 15N enrichment of soil at 50 cm relative to surface litter and the strength of the regression increases to an adjusted r2 of 0.46 (Table 4).

display math(Eqn 9)
Table 4. Regressions of climate and mycorrhizal type against 15N enrichment between surface litter and deep soil
ParameterValue ± SE P
  1. Climate, soil data, and mycorrhizal type are from Hobbie & Ouimette (2009). Mycorrhizal type is based on dominant vegetation, and classified as arbuscular mycorrhizal (AM) (32), ectomycorrhizal (ECM) (47), and mixed arbuscular/ectomycorrhizal (6) (Supporting Information Table S2). MAT, mean annual temperature (°C); MAP, precipitation (mm).

Adjusted r20.456< 0.001
Intercept9.1 ± 1.1< 0.001
MAT−0.028 ± 0.0570.624
MAP3.1 ± 2.6e-40.222
Mycorrhizal type< 0.001
AM−3.29 ± 0.51< 0.001
ECM1.18 ± 0.550.037

Thus, 15N enrichment in soil profiles is c. 4.5‰ higher in systems dominated by ectomycorrhizal symbioses than in systems dominated by arbuscular mycorrhizal symbioses. Transfer of 15N-depleted N to ectomycorrhizal plants and retention of 15N-enriched N by ectomycorrhizal fungi appears to drive 15N depletion in surficial litter layers and 15N enrichment in deeper soil horizons, as also suggested in several previous studies (Högberg et al., 1996; Billings & Richter, 2006; Lindahl et al., 2007; Hobbie & Ouimette, 2009). For example, in chronosequences of boreal forests the lowest part of the surficial organic mor layer becomes increasingly enriched in 15N with age (Wallander et al., 2009). In these systems, correlations of 15N with fungal biomass and low pH (limiting nitrification) led the authors to conclude that redistribution of N isotopes resulted from the ectomycorrhizal symbiosis rather than from N losses.

3. Forms of nitrogen

Although isotopic measurements of bulk N are common, measurements of other N forms are scarce. Ammonification discriminates slightly against 15N (Koba et al., 2003), whereas the discrimination against 15N during nitrification and denitrification is larger (Table 1). The relative fluxes of these three processes will accordingly affect the isotopic relationships among bulk soil, ammonium, and nitrate (Houlton et al., 2007). Nitrate is usually 1–6‰ depleted in 15N relative to ammonium in N-limited systems. Such patterns have been reported from coniferous forests (Choi et al., 2005; Takebayashi et al., 2010) and Arctic tundra (Yano et al., 2009). Significant nitrification can lead to 15N enrichment of ammonium relative to the bulk soil and significant further denitrification can lead to 15N enrichment of nitrate relative to ammonium (Koba et al., 1998; Houlton et al., 2007).

The δ15N of organic forms of N are sometimes measured in soil. In general, dissolved organic N appears similar to bulk N values (Houlton et al., 2007), although Yano et al. (2009) reported low δ15N for dissolved organic N relative to other measured pools in Arctic tundra. According to models and data across soil density gradients (Sollins et al., 2006, 2009), dense, mineral-associated organic N that derived from proteinaceous microbial residues was particularly high in 15N, with a maximum enrichment of 4.8‰ between the lightest and the heaviest mineral fractions (< 1.65 to > 2.55 g cm−3).

IV. Patterns of fungal δ15N

Mayor et al. (2009) compiled the available isotopic data on fungi and determined that little relationship existed between climatic variables and N isotope patterns in saprotrophic and ectomycorrhizal fungi. Ectomycorrhizal fungi averaged 6.7‰ higher in 15N than saprotrophic fungi. Other studies have reported that litter decay fungi are 1.6–2.4‰ higher in 15N than wood decay fungi (Kohzu et al., 1999; Hobbie et al., 2001). Some of the 15N enrichment of ectomycorrhizal fungi relative to saprotrophic fungi undoubtedly arises because saprotrophic fungi are active higher in the soil profile than ectomycorrhizal fungi (Lindahl et al., 2007), but physiological differences, particularly the transfer of 15N-depleted N from mycorrhizal fungi to host plants (Hobbie & Colpaert, 2003), are also important.

1. Patterns among taxa

Functional attributes may correlate with N isotope patterns in ectomycorrhizal fungi. Lilleskov et al. (2002) and Trudell et al. (2004) suggested that rhizomorph abundance and δ15N could be linked, with taxa with thick rhizomorphs (aggregated hyphae for long-distance transport) such as Cortinarius and Tricholoma generally higher in δ15N than other taxa. Hobbie & Agerer (2010) analyzed data from two previous studies of sporocarp isotope patterns (Taylor et al., 2003; Trudell et al., 2004) to confirm that N isotopes correlated with functional attributes of ectomycorrhizal fungi; specifically with how they explore the soil (termed exploration type; Agerer, 2006) and with the hydrophobicity of ectomycorrhizas. The ectomycorrhizal exploration types separated into two groups isotopically, with exploration types with hydrophobic ectomycorrhizas averaging 3–4‰ higher in 15N than exploration types with hydrophilic mycorrhizas. In Table 5 we have used the data of Mayor et al. (2009) to report average values for ectomycorrhizal fungi by genus, hydrophobicity, and exploration type.

Table 5. δ15N for common ectomycorrhizal fungi
GenusMean (‰) n HydrophobicityaExploration typea
  1. Values are ± SE. Nitrogen represents the number of distinct species × site combinations. Global averages are derived from Mayor et al. (2009) except Tuber values which are from Hobbie et al. (2001).

  2. a

    Fungal exploration type and presence of hydrophilic or hydrophobic ectomycorrhizas after Agerer (2006).

  3. b

    Ammonophilic taxa, data from Kohzu et al. (1999), classifications from Imamura & Yumoto (2008).

Laccaria 0.5 ± 0.615HydrophilicShort
Inocybe 2.3 ± 0.424HydrophilicShort
Hebeloma 2.7 ± 1.17HydrophobicShort/medium-fringe
Amanita 3.1 ± 0.535HydrophilicMedium-smooth
Russula 3.2 ± 0.383HydrophilicContact/short/medium-smooth
Lactarius 4.2 ± 0.354HydrophilicContact/medium-smooth
Dermocybe 4.9 ± 1.411HydrophobicMedium-fringe
Boletus 5.8 ± 1.017HydrophobicLong
Cortinarius 6.8 ± 0.3100HydrophobicMedium-fringe
Paxillus 7.1 ± 0.77HydrophobicLong
Leccinum 8.1 ± 0.719HydrophobicLong
Suillus 8.2 ± 0.717HydrophobicLong
Tricholoma 9.3 ± 0.635HydrophobicMedium-fringe
Hydnellum 9.4 ± 1.06HydrophobicMedium-mat
Tuber 15.1 ± 0.69HydrophilicShort
Hebeloma c. 212HydrophobicShort/medium-fringeb

Because soil δ15N increases with increasing depth, the depth at which taxa obtain their N should also correlate with δ15N values. Some general patterns in depth distributions of fungal activity have emerged from studies using morphological or genetic characteristics to assess fungal identities of hyphae or mycorrhizas in the soil profile (Landeweert et al., 2003; Rosling et al., 2003; Scattolin et al., 2008). A rigorous comparison between exploration depth and δ15N has not yet been carried out, although Agerer et al. (2012) reported that δ15N of Ramaria correlated with the observed depth of hyphal exploration. In two Swedish studies, hyphae harvested from in-growth bags increased by 1.6‰ from 5 to 10 cm depth (Boström et al., 2007), increased by 3.5‰ from 5 to 15 cm, and increased by 4.6‰ from 5 to 25 cm depth (Wallander et al., 2004). Similarly, ectomycorrhizal root tips increased by 5‰ from the uppermost mor layer to 0–5 cm in the mineral soil (Högberg et al., 1996).

High δ15N values pose particular difficulties in interpretation, as they are generally higher than any concurrently measured bulk soil pool that could serve as a source. In comparing patterns in δ15N of soil horizons and the colonization by horizon of different ectomycorrhizal taxa (Lindahl et al., 2007) with the δ15N of those taxa collected from a nearby site (Taylor et al., 2003), ectomycorrhizal fungi appear to be 5–9‰ enriched in 15N relative to the soil horizons from which they assimilate N (Table 6). The limited data from this study confirm that ectomycorrhizal fungi are active deeper in the soil profile than saprotrophic fungi, with a 2.8‰ difference in δ15N between the potential N sources for ectomycorrhizal fungi (−1.1‰) and saprotrophic fungi (−3.9‰), with two ericoid mycorrhizal taxa intermediate at −1.9‰. With overall averages for Stadsskogen fungi at 5.8‰ for ectomycorrhizal fungi and 0.8‰ for saprotrophic fungi, this translates into a 6.9‰ enrichment in 15N for ectomycorrhizal fungi and a 4.7‰ enrichment in 15N for saprotrophic fungi relative to the soil horizons where these fungi are present.

Table 6. Estimated δ15N signature (±SE) of source nitrogen (N) for different taxa, based on colonization patterns of taxa in different soil horizons and the δ15N values of those horizons (Lindahl et al., 2007)
TaxaSourceδ15N (‰)Sporocarpδ15N (‰)TaxaSourceδ15N (‰)
  1. Data are given in Supporting Information Table S3.

EctomycorrhizalEricoid mycorrhizal
Cortinarius spp. (= 5)−2.2 ± 0.56.5Capronia sp.−2.1
C. armeniacus −2.55.4 Rhizoscyphus ericae −1.6
C. armeratus −0.6 Saprotrophic
Dermocybe sp.−3.7  Galerina atkinsoniana −3.5
D. semisanguinea −2.56.4 Lophodermium −4.2
Hydnellum aurantiacum 1.610.6 L. pinastri −3.2
Piloderma sp.2.4  Marasmius androsaceus −4.6
P. reticulatum −0.9 Mycena sp.−4.0
Rozites caperatus 1.06.6Averages
Sebacina sp.−0.5 Ectomycorrhizal−1.1 ± 0.5
Thelephorales sp.0.5 Ericoid mycorrhizal−1.85 ± 0.25
Phialocephala fortinii −0.6 Saprotrophic−3.9 ± 0.2

Many taxa with high δ15N values also possess strong proteolytic ability (Lilleskov et al., 2002), so high δ15N may be a marker of organic N uptake (e.g. Cortinarius). In addition, two Hebeloma species collected by Kohzu et al. (1999) in Okinawa and Japan are classified as ammonophilic fungi (Imamura & Yumoto, 2008), with fruiting often indicating the presence of dead animals or rodent burrows. Thus, the very high values for these two taxa (c. 20‰) presumably indicate assimilation of 15N-enriched ammonium after ammonia volatilization. Relative to other ectomycorrhizal taxa, δ15N values for Tuber are also very high. However, Tuber spp. primarily colonize the mineral soil (Baier et al., 2006; Buée et al., 2007; Scattolin et al., 2008), where the high δ15N typical of deep mineral soil horizons (Hobbie & Ouimette, 2009) therefore probably contributes to the high δ15N value of Tuber.

2. Patterns within fungi

Isotopic patterns in different fungal components may provide some insight into mechanisms creating isotopic differences among fungi. Taylor et al. (1997) reported that protein and amino acids were c. 10‰ enriched in 15N relative to chitin in fungi and also suggested that the higher 15N abundance and %N in caps relative to stipes were related to more protein and amino acids in caps than in stipes. This 15N enrichment probably accounts for other reported patterns within fungi, including 15N enrichment of fungal tissue in ectomycorrhizas relative to extraradical hyphae (Hobbie & Colpaert, 2003) and 15N enrichment in sporocarps relative to extraradical hyphae (Kohzu et al., 2000; Wallander et al., 2004; Boström et al., 2007; Table 7).

Table 7. 15N and %N difference (designated ε) in culture studies and field studies in ectomycorrhizal fungi and plants
Tissues comparedε (15N (‰))ε (%N)References
  1. a

    Ammonium nutrition; Pinus sylvestris with Suillus bovinus.

  2. b

    Nitrate nutrition.

  3. c

    Culture study with Suillus variegatus and Pinus densiflora.

  4. d

    Mycelia from 0 to 10, 10 to 20, and 20 to 30 cm in spruce and mixed species stands.

  5. e

    Sporocarps for seven ectomycorrhizal taxa averaged 6.3‰; mycelia from in-growth bags averaged 3.8 ± 0.5 at 5 cm and 5.4 ± 1.5‰ at 10 cm (± SE). Stand is Picea abies.

  6. f

    Hydrophobic ectomycorrhizas.

  7. g

    Hydrophilic ectomycorrhizas.

  8. h

    Tricholoma sciodes.

Culture studies
Rhizomorphs – shoots3–5a, 0–3bHögberg et al. (1999)
Sporocarps – mycelia2.7cKohzu et al. (2000)
Field studies
Sporocarps – mycelia2.8d2.0%Wallander et al. (2004)
Sporocarps – mycelia0.3, 6.8Clemmensen et al. (2006)
Sporocarps – mycelia0.9, 2.5eBoström et al. (2007)
Caps – stipes (±SE)2.1 ± 0.82.0 ± 1.0Taylor et al. (1997)
Caps – stipes (ho)f3.21.9Hobbie et al. (2012)
Caps – stipes (hi)g1.41.5Hobbie et al. (2012)
Gills – stipeshc. 6Zeller et al. (2007)

V. Biochemical basis for the influence of fungi on δ15N patterns in plant–soil systems

Kinetic isotopic effects associated with movement of ammonia or amino groups within mycorrhizal fungi and the subsequent transfer to host plants of ammonia or amino acids appear to deplete host plants in 15N relative to fungal symbionts. This process would be analogous to that presumably causing 15N enrichment of up to 10‰ in N2-fixing root nodules relative to soybean (Glycine max) leaves (Shearer et al., 1980; Kohl et al., 1983; Fig. 2), in which ammonia is the transfer compound. Such processes would also enrich mycorrhizal fungi in 15N relative to saprotrophic fungi, even if source N were similar for these two fungal types. Ammonia is a suspected transfer compound in arbuscular mycorrhizal fungi (Bago et al., 2001), whereas glutamine, glutamate, alanine, and ammonia have all been invoked as potential transfer compounds in ectomycorrhizal fungi (Smith & Smith, 1990; Chalot et al., 2006; Dietz et al., 2011). In the hypothesized glutamine–glutamate shuttle mechanism for N transfer between ectomycorrhizal fungi and plants, the amido group of glutamine may be the source N assimilated by ectomycorrhizal plants (Smith & Smith, 1990), and is the source for N in N-acetylglucosamine (Zalkin & Smith, 2006), the monomer of chitin. Thus, 15N depletion of chitin relative to co-occurring protein parallels the 15N depletion of compounds that are subsequently transferred to host plants. The 15N depletion of chitin appears to be a general phenomenon accompanying chitin biosynthesis, as chitin in arthropods, marine invertebrates, or fungi is depleted in 15N relative to muscle, total biomass, or protein by 7–12‰ (Schimmelman & DeNiro, 1986; Macko et al., 1989; Taylor et al., 1997; Webb et al., 1998). Recent work estimated that the 15N enrichment of protein relative to chitin averaged 15‰ for hydrophobic ectomycorrhizal fungi, 10‰ for saprotrophic fungi, and only 7‰ for hydrophilic ectomycorrhizal fungi (Hobbie et al., 2012), with variable isotopic partitioning within a closed system (e.g. Eqn (Eqn 6)) as a plausible mechanism to create these patterns.

Figure 2.

Enrichment in 15N of root nodules relative to leaves vs the transfer efficiency of nitrogen (N) from root nodules to the rest of the plant. Data are from studies in soybean (Glycine max) by Shearer et al. (1980) (closed symbols) and Kohl et al. (1983) (open symbols).

VI. Patterns of δ15N in plant and fungal culture studies

To better understand how mycorrhizal fungi influence plant δ15N, several studies have grown plants in symbiosis with arbuscular mycorrhizal or ectomycorrhizal fungi with different N forms and reported the resulting isotopic patterns. The few studies with arbuscular mycorrhizal plants have been difficult to interpret, with no clear indication that arbuscular mycorrhizal fungi are sequestering a 15N-enriched pool or passing 15N-depleted N to host plants (Azcón-G-Aguilar et al., 1998; Wheeler et al., 2000). However, fractionation on uptake cannot be ruled out in these studies, and it is difficult to quantify N retention by arbuscular mycorrhizal fungi. Studies with ectomycorrhizal plants have most commonly used Pinus sylvestris. Relative to nonmycorrhizal plants, foliar δ15N declined 0.5–4.6‰ with mycorrhizal colonization, with smaller declines under nitrate supply than under ammonium or ammonium nitrate supply (Table 8). Many ectomycorrhizal fungi appear to assimilate nitrate more slowly than ammonium (Finlay et al., 1992; Keller, 1996), with some transfer of unreduced nitrate to host roots (Ek et al., 1994; Hobbie et al., 2008). The transfer of 15N-depleted N from fungi to plants coupled with N retention by fungi leads to 15N-depleted plants and 15N-enriched fungi. The 15N patterns can therefore provide insight into N partitioning between mycorrhizal fungi and host plants, and also appear to correlate highly with fungal biomass (Fig. 3).

Figure 3.

Fungal biomass correlates strongly with foliar δ15N in mycorrhizal pine (r2 = 0.88; P < 0.001). Nitrogen (N) as ammonium nitrate was supplied exponentially at rates of 3% d−1 (low N; open symbols) or 5% d−1 (high N; closed symbols) to Pinus sylvestris cultured with either Thelephora terrestris (squares) or Suillus luteus (triangles). Nitrogen was the limiting nutrient in both treatments. Data are from Hobbie & Colpaert (2003).

Table 8. 15N depletion in foliage of mycorrhizal pine vs nonmycorrhizal pine, calculated as δ15Nnonmycorrhizal − δ15Nmycorrhizal in per mil (‰)
Species/symbiontNitrogen formReferences
NH4NO3NH4+NO3
  1. a

    Unknown ectomycorrhizal symbionts.

  2. b

    Symbiont, Suillus variegatus.

  3. c

    Low nitrogen (N) supply rate.

  4. d

    High N supply rate.

Pinus halepensis a 2.0Bardin et al. (1977)
Pinus densiflora b 1.7–4.6Kohzu et al. (2000)
Pinus sylvestris
 Suillus luteus 0.5c, 2.8dHobbie & Colpaert (2003)
 Thelephora terrestris 0.9c, 1.9dHobbie & Colpaert (2003)
 Suillus bovinus 3.5c, 1.7d0.5c, 1.3dHobbie et al. (2008)
 Laccaria laccata 2.3c, 3.0d2.2c, 1.4dHobbie et al. (2008)

Only two isotopic studies have cultured mycorrhizal plants with organic N forms. In ectomycorrhizal Betula nana and ericoid mycorrhizal Vaccinium vitis-idaea, only a small proportion of supplied glutamic acid and glycine was assimilated, with 15N fractionation on uptake of at least 5‰ for glutamine, glutamic acid, and glycine (Emmerton et al., 2001a). Similarly, in ectomycorrhizal Eucalyptus supplied with glutamine, glutathione, or bovine serine albumin (BSA), < 30% of supplied N was assimilated (estimated graphically; Schmidt et al., 2006). Fractionation for BSA appeared negligible, whereas Eucalyptus was actually enriched in 15N relative to the supplied glutathione. To assess whether internal 15N redistributions in the plant–mycorrhizal system differ, further culture studies with organic N compounds are needed where fungal N retention is quantified and where all supplied N is assimilated.

VII. Mycoheterotrophic and parasitic plants

Mycoheterotrophic plants lack chlorophyll and depend on specific mycorrhizal fungi for supplies of C and N (Leake, 1994). Many of the 400 species are orchids, but some are monotropoid and pyroloid plants (Gebauer & Meyer, 2003; Zimmer et al., 2007). Fully mycoheterotrophic plants were c. 12‰ higher in 15N than autotrophic reference plants (Zimmer et al., 2007), and mycoheterotrophic plants were c. 3–4‰ higher in 15N than their associated ectomycorrhizal fungi (Trudell et al., 2003). Based on consistent 15N enrichment in protein and amino acids relative to chitin in other studies, Trudell et al. (2003) suggested that the high δ15N of mycoheterotrophic plants relative to possible fungal symbionts reflected preferential incorporation of protein-derived N vs chitin-derived N. Mycoheterotrophic plants may acquire N by digestion of fungal structures (Smith & Read, 2008), which presumably differs sufficiently from the controlled exchange between mycorrhizal fungi and autotrophic plants for different isotopic patterns to arise. An isotopic mass balance between the 3 and 4‰ enrichment in 15N together with a 15N enrichment of protein relative to chitin of perhaps 10‰ suggests that fungal N is 30–40% chitin. More recently, δ15N patterns in mycoheterotrophic plants associated with either arbuscular mycorrhizal fungi or saprotrophic fungi have been investigated, with mycoheterotrophic plants associated with arbuscular mycorrhizal fungi similar in δ15N to their fungal hosts (Merckx et al., 2010), whereas mycoheterotrophic plants associated with saprotrophic fungi were c. 3‰ depleted in 15N relative to their fungal hosts (Martos et al., 2009; Ogura-Tsujita et al., 2009).

Partial mycoheterotrophy in photosynthetic orchids and pyroloid plants has also been investigated (Gebauer & Meyer, 2003; Hynson et al., 2009). The degree of dependence on heterotrophy for N and C acquisition in partial mycoheterotrophs can be estimated using an isotopic mixing-model analysis with co-existing autotrophs and full mycoheterotrophs as isotopic endmembers. Variation in the δ15N and δ13C among autotrophs, mycoheterotrophs, and partial mycoheterotrophs may accordingly reveal information not only about the sources of N and C, but about the C cost for N uptake in partial mycoheterotrophs.

Parasitic plants that tap into the xylem or phloem of other plants for their N supply provide an interesting isotopic contrast to mycoheterotrophic plants. Autotrophic plant hosts lack the 15N-depleted chitin that can strongly influence internal partitioning of 15N in fungi and, as a consequence, parasitic plants appear similar in δ15N to their plant hosts (within 1‰). Such isotopic measurements were used to determine that Santalum acuminatum primarily obtained N from co-occurring N2-fixing legumes in Australian heathlands (Tennakoon et al., 1997) and to determine that Geocaulon lividum tapped into the canopy dominants in boreal Alaska (Hobbie et al., 2009).

VIII. Patterns of foliar δ15N in autotrophic plants

1. Global patterns of foliar δ15N in non-N2-fixing autotrophic plants

Researchers have published several compilations of data on foliar δ15N across biomes since 1997. In one survey, foliage from tropical forests averaged 6.5‰ higher than foliage from temperate forests, and tropical soils averaged 8‰ higher than temperate soils (Martinelli et al., 1999). The authors attributed this to a greater availability of N in the tropical systems (which are often limited by phosphorus rather than by N), and hence, larger losses of N through processes that discriminate against 15N. The latter suggestion has been supported by modeling of denitrification in montane tropical rainforests (Houlton et al., 2006). Across a wide range of ecosystems, site-averaged foliar δ15N was negatively correlated with mean annual precipitation, with ecosystems distant from the equator and with high rainfall having particularly low site-averaged foliar δ15N (Handley et al., 1999). The available evidence suggests that plant δ15N broadly reflects soil δ15N, so low δ15N values should reflect the low importance of loss mechanisms of 15N-depleted N2, N2O, and nitrate from soil.

The most complete compilation of data on foliar N isotopes (including 11 000 plants world-wide) was reported by Craine et al. (2009). For non-N2-fixing plants, foliar δ15N decreased as precipitation increased (for mean annual temperature > −0.5°C). Importantly, the type of mycorrhiza formed by plants influenced δ15N, with foliar δ15N decreasing in the order nonmycorrhizal (mean ± SE, 0.9 ± 0.2‰) > arbuscular mycorrhizal (−1.1 ± 0.1‰) > ectomycorrhizal (−2.3 ± 0.2‰) > ericoid mycorrhizal plants (−5.0 ± 0.2‰). These values were normalized to a standard value for temperature (13.2°C), precipitation (751 mm yr−1), and N concentration (1.58%). In the database regression of foliar δ15N, 56% of variance could be explained. Of that variance, precipitation explained 14%, temperature and derived variables an additional 14%, mycorrhizal association 29%, and foliar N concentration 44%. Overall, nonmycorrhizal plants were enriched in 15N relative to all mycorrhizal types, suggesting either that the source N for nonmycorrhizal plants was enriched in 15N relative to source N for mycorrhizal plants, or that fractionation during creation of transfer compounds by mycorrhizal fungi resulted in mycorrhizal plants receiving a 15N-depleted N pool. The importance of foliar N concentration as an explanatory variable might reflect a dichotomy between N dynamics in systems with plants of deciduous vs evergreen foliage, but could also reflect correlations among N availability, foliar N concentrations, belowground allocation, relative C allocation to mycorrhizal fungi, and retention of 15N-enriched N belowground (Ågren & Bosatta, 1996; Hobbie, 2006).

These global patterns overestimate isotopic differences among plants of different mycorrhizal type when they co-occur. This is particularly apparent when comparing ericoid mycorrhizal plants against ectomycorrhizal and arbuscular mycorrhizal plants, where site-specific differences are 2‰ less than overall average differences, as shown in Table 9. This pattern presumably arises because N dynamics causing high δ15N in ectomycorrhizal and arbuscular mycorrhizal plants (generally, high rates of N cycling) tend to exclude ericoid mycorrhizal plants.

Table 9. Plant δ15N of differing mycorrhizal types were compared at sites where they co-occur (with paired t-tests)
Mycorrhizal types comparedSite-averaged 15N difference (‰) P n Overall 15N difference (‰)
  1. Data are from the Craine et al. (2009) database. The number of sites where the specified mycorrhizal types co-occur is given by n. Climate-adjusted overall δ15N averages as reported in Craine et al. (2009) are also given. AM, arbuscular mycorrhizal; ECM, ectomycorrhizal; ERM, ericoid mycorrhizal; Non, nonmycorrhizal.

AM – ECM0.70.215181.2
AM – ERM1.80.002143.9
Non – AM2.30.001212.0
ECM – ERM0.70.140102.7
Non – ECM3.90.00593.2
Non – ERM4.60.00275.9

2. Site-specific foliar 15N patterns and the influence of N availability

Many site-specific studies have reported that ectomycorrhizal and ericoid mycorrhizal plants in Arctic, alpine or boreal regions were significantly depleted in 15N relative to co-occurring arbuscular mycorrhizal plants (Schulze et al., 1994; Michelsen et al., 1996, 1998; Hobbie et al., 2005). The very high biomass of ectomycorrhizal fungi that may accompany the N-limited conditions prevalent in these regions could potentially sequester sufficient 15N-enriched N to deplete in 15N the remaining pool of N available for plant transfer. Alternatively, the extensive enzymatic capabilities of ericoid mycorrhizal and ectomycorrhizal fungi may allow their host plants to access fresh, litter-derived N, whereas arbuscular mycorrhizal fungi (and plants) must rely on deeper N that has been enriched in 15N during microbial processing. Comparing tracer 15N labeling studies to natural abundance δ15N patterns to distinguish among N sources from different depths could assess these two competing hypotheses.

Numerous studies have linked changes in foliar δ15N across sites or across time to changes in N availability. Declines over time in foliar δ15N and %N in herbarium specimens have been attributed to increasing N limitation caused by the c. 100 ppm rise in atmospheric CO2 over the last 150 yr (Peñuelas & Estiarte, 1997; McLauchlan et al., 2010). Declines (averaging 1‰ in 27 species) in foliar δ15N in free-air CO2 enrichment (FACE) studies were attributed by the authors to either interactions with mycorrhizal fungi or increased assimilation of nitrate in roots (Bassirirad et al., 2003). In a Pinus strobus chronosequence in northeastern North America, Compton et al. (2007) used an N isotope mass balance approach to conclude that N cycling became tighter during 100 yr of stand development, during which foliar δ15N declined. Declines of 1‰ in δ15N of tree rings and lake sediments also indicated decreased N availability over the last 75 yr of forest development in the northeastern USA (McLauchlan et al., 2007). Such analyses should be restricted to heartwood, as comparing data from old heartwood vs young sapwood could be easily confounded by differences in metabolic activity.

Long-term records avoid some of the issues with interpreting retrospective analyses such as tree rings or sediment cores. In a study of a boreal pine forest, the δ15N of needles collected over a period of 35 yr fell by c. 3‰ in the control plots (Högberg et al., 2011); in heavily N-fertilized plots the δ15N of needles increased, but declined after the treatment ended. Strong correlations among the δ15N of needles, DNA sequences of ectomycorrhizal fungi and a phospholipid fatty acid biomarker attributed to ectomycorrhizal fungi confirmed the role of ectomycorrhizal fungi in determining the δ15N of the plants. In Hietz et al. (2011), increased δ15N in trees in tropical forests in Panama (leaves) and Thailand (wood) over a 40- to 80-yr period was attributed to anthropogenic N deposition increasing N losses from these ecosystems. After a ploughed agricultural field was planted with ectomycorrhizal pine trees in South Carolina, a 15‰ difference developed over 40 yr between the upper organic layer, which derives from recent aboveground plant litter, and the mineral soil at 35–60 cm depth (Billings & Richter, 2006). The 15N enrichment at depth was attributed to accumulation of 15N-enriched N derived from microbes, including ectomycorrhizal fungi.

3. Using plant δ15N to estimate carbon flux to mycorrhizal fungi

Because N and C cycling are closely coupled, it is possible to use stoichiometric relationships to mathematically link them. Such stoichiometry is particularly important in mycorrhizal fungi, where the potential N gain from exploring soil must be balanced against both the C cost and the N cost (for chitin and protein) of building the necessary tissues to extend into previously unexploited regions. Under some conditions, plant δ15N values in mycorrhizal systems may reflect the balance of N between the plant and fungal components of the system (Fig. 1 and Eqns (Eqn 4) and (Eqn 5)), with the mycorrhizal transfer ratio Tr (the proportion of N taken up by fungi that is subsequently transferred to plant hosts) as the controlling variable. In ectomycorrhizal culture, C allocation to fungi correlates strongly and negatively with plant δ15N values (Fig. 3). This C allocation can be mathematically expressed as a function of Tr, the plant N supply from mycorrhizal fungi (Np), the C/N of fungal biomass, and the efficiency (e) with which plant-supplied C is turned into fungal biomass (Hobbie & Hobbie, 2008).

display math(Eqn 10)

The mycorrhizal transfer ratio reflects the C cost to acquire N by mycorrhizal fungi, and is accordingly the key variable determining both Cfungal and plant δ15N. This is illustrated in Fig. 4, which expresses the relationship between Tr, plant δ15N, and the plant C cost to acquire a unit of mycorrhizally derived N (Cfungal/Np). The C cost of N acquisition increases as Tr declines, such as under conditions of low N availability in many ectomycorrhizal forests. Hobbie & Hobbie (2006) used equations similar to Eqn (Eqn 10) and system N isotope patterns to estimate that 8–17% of plant productivity was allocated to mycorrhizal symbionts in a tundra shrub ecosystem. Related equations were used to estimate the fraction of plant N supplied by mycorrhizal fungi.

Figure 4.

Carbon (C) cost for nitrogen (N) supply from mycorrhizal fungi to plants (Cfungal/Np) increases as the mycorrhizal transfer ratio (Tr) for N decreases, according to Eqn (Eqn 10). Fungal C : N is set to 10 and microbial efficiency (e) is set to 0.5. Tr is the proportion of N taken up by fungi that is transferred to host plants. All plant N is assumed to be supplied by mycorrhizal fungi. Variation in plant δ15N as a function of Tr is also shown, with 15N fractionation during creation of N transfer compounds by mycorrhizal fungi (Δf) set at 10‰.

IX. Controls over plant δ15N

Numerous explanations have been invoked to explain N isotope patterns in plants. One difficulty is that several processes can plausibly account for a given set of isotopic patterns, and it is tempting to explain a specific isotopic pattern by claiming that an unmeasured process is the likely control. Here, we discuss some of the most plausible explanations for patterns of plant δ15N.

1. Rooting depth

Because soils commonly increase in δ15N with depth, differences in δ15N are sometimes attributed to differences in rooting depth across species or, less commonly, within a species and across sites. For example, Kohzu et al. (2003) correlated rooting depth in different species of mire plants against foliar δ15N, and Hobbie et al. (2009) attributed very low δ15N in Picea mariana in Alaska to its association with permafrost, which would restrict foraging to the uppermost litter and soil layers. Unfortunately, little information exists on rooting distributions with depth for most species. An additional difficulty is that rooting distributions are unlikely to correspond closely with uptake depths (as shown for N and other elements by Göransson et al., 2006, 2008), particularly for elements like N whose concentration often declines swiftly with depth.

2. Source differences

Uptake of different forms of N could affect plant δ15N. Ammonium and nitrate are the N forms commonly measured, although a few studies have measured forms of organic N, such as dissolved organic N (Takebayashi et al., 2010), hydrolyzed amino acids (Bol et al., 2004) or amino sugars (Yano et al., 2009). Because nitrification and denitrification both discriminate against 15N, the relative importance of these two processes controls whether nitrate is higher or lower in δ15N than ammonium.

Several studies have compared plant δ15N against measures of nitrate vs ammonium use. In Swedish oak woodlands (Falkengren-Grerup et al., 2004) and in alpine tundra (Miller & Bowman, 2002), greater relative nitrate uptake correlated negatively with foliar δ15N in herbaceous plants, suggesting δ15Nnitrate < δ15Nammonium, whereas German temperate grasslands had the reverse pattern, leading Kahmen et al. (2008) to conclude that δ15Nnitrate > δ15Nammonium. Denitrification is presumably higher where δ15Nnitrate > δ15Nammonium.

The influence of mycorrhizal fungi on the sources of N available to plants and on the isotopic signature of the N derived from those sources that are ultimately transferred to plants makes estimating the importance of ammonium, nitrate, and organic N as sources very difficult. At sufficiently high concentrations, fractionation against 15N on uptake is an additional factor that is difficult to quantify under field conditions. One possible solution to assess the δ15N of plant-available N is to use buried cellulose filters as integrators of microbially available N (Hendricks et al., 1997). The cellulose serves as a relatively labile C source for colonizing microbes, which then assimilate available N from the soil. In a 15N labeling study, this approach correlated better with plant δ15N than using either inorganic N extracted from resin bags or buried bag incubations (Hendricks et al., 2004). Solid samples like cellulose filters are also easier to process than aqueous ions such as ammonium and nitrate, and an integrated signal of available N (including organic N) is assessed, rather than only inorganic N.

3. Fractionation on uptake

Substantial fractionation against 15N on uptake and assimilation is possible if N concentrations are high relative to uptake rates (Handley & Raven, 1992; Fogel & Cifuentes, 1993; Emmerton et al., 2001b). This can be an issue in culture studies, where applied N concentrations are generally higher than in field situations. Therefore, unless N supply is carefully matched to uptake rates, δ15N patterns in culture have uncertain relevance to field situations, where N supply rates appear generally low. Fractionation on uptake depends on the medium, with diffusion rates and fractionation higher in liquid culture than in agar culture (reviewed in Hobbie & Hobbie, 2008), and also depends on the form of N (generally, diffusion of nitrate > ammonium > organic N).

Fractionation on uptake is generally discounted as a controlling factor for plant δ15N in field studies, under the assumption that N concentrations at the site of uptake are too low for fractionation to be expressed. However, phosphorus limitation may encourage fractionation against 15N on assimilation (Högberg et al., 1999), as seen in mangroves (McKee et al., 2002) and bogs (Clarkson et al., 2005). Sites with high water tables may also facilitate fractionation on uptake (Kohzu et al., 2003). Field conditions that encourage fractionation against 15N during uptake (such as higher inorganic N concentrations) also favor losses of 15N-depleted N through ammonia volatilization, nitrate leaching or denitrification, thereby increasing the δ15N of the remaining N.

4. Mycorrhizal fungi

Mycorrhizal fungi can influence plant δ15N in several ways. By increasing plant access to recalcitrant and slowly diffusible forms of N, they may alter the average δ15N of the available N sources. This could include 15N-depleted sources in surficial litter and 15N-enriched sources at greater depths. Their increased surface area and uptake capacities at low external concentrations may also decrease the extent of fractionation on uptake. Finally, as previously discussed, biochemical reactions within fungi may partition N into 15N-enriched and 15N-depleted pools, with plants apparently receiving N primarily from the latter pool.

X. Conclusions and research needs

From the analyses presented in this review, we propose that interactions among plants, mycorrhizal fungi, and soil drive many ecosystem δ15N patterns, with the type of mycorrhizal association correlating strongly with values in bulk soil, soil profiles, and foliage. By contrast, climatic factors explained little of the observed variance once mycorrhizal associations were accounted for. Although including mycorrhizal fungi in large-scale databases of soil δ15N increased the fraction of variance explained, overall ability to predict bulk soil δ15N from climate and mycorrhizal association remained rather poor, indicating that large-scale patterns presumably reflect factors not captured in current statistical models. For foliar δ15N, the large fraction of variance explained by N concentrations should be investigated further. High N availability can increase foliar %N while increasing losses of 15N-depleted N and raising the δ15N of available N. In addition, high N availability could diminish sequestration of 15N-enriched N in fungal biomass and accordingly diminish the 15N fractionation mediated by mycorrhizal fungi between available N and plants.

Colonization by ectomycorrhizal and ericoid mycorrhizal fungi increases plant access to N forms that are poorly available to nonmycorrhizal plants. This N may be 15N-depleted (e.g. litter N) or 15N-enriched (e.g. mineral horizons) relative to that available to nonmycorrhizal plants, but determining the relative contribution from different soil N pools to plant N budgets is not yet generally possible from natural abundance measurements. Low δ15N in ericoid mycorrhizal plants relative to co-occurring ectomycorrhizal and arbuscular mycorrhizal plants probably arises from ericoid mycorrhizal fungi mainly acquiring N from 15N-depleted litter layers. Consequently, N acquired by ericoid mycorrhizal fungi and plants may be of shallower and more recent origin than N acquired by other plant types.

15N fractionation during creation of transfer compounds by mycorrhizal fungi leads to retention of 15N-enriched N and transfer of 15N-depleted N to the plant symbiont. This 15N fractionation probably occurs in all autotrophic mycorrhizal plants and makes it difficult to directly link plant δ15N to soil N sources. Transfer compounds may be ammonia or amino acids; 15N fractionation during amino acid synthesis, ammonia formation from urea, or aminotransferase reactions such as formation of N-acetylglucosamine are possible. Ectomycorrhizal fungi vary widely in their enzymatic capabilities; fungal δ15N corresponds to the N source (e.g. δ15Ninorganic < δ15Norganic, usually), the depth of N acquisition (δ15Nshallow < δ15Ndeep), and the mycorrhizal transfer ratio, although more work is needed on these factors. Sporocarp δ15N probably reflects closed-system rather than open-system isotopic fractionation. Partitioning of N between 15N-enriched fungal protein and 15N-depleted fungal chitin appears to be another key control over plant, fungal, and soil δ15N patterns.

Ecosystem-scale modeling of 15N distributions over time at natural abundance using well-characterized systems (Hobbie et al., 1999; Billings & Richter, 2006) can help to constrain the range of possible scenarios that fit the available data. As one example, the Non-Equilibrium Stable Isotope Simulator (NESIS) model can be used to take output from an element-based model and then predict isotopic signatures for the pools and fluxes in that model (Rastetter et al., 2005).

In another approach, 15N labeling can be followed for multiple years to further constrain the range of possible solutions (Currie & Nadelhoffer, 1999). This approach was used to determine that N from mineral soil was an important N source for trees (Currie et al., 2004). When combined with natural abundance δ15N data, 15N labeling could further help to distinguish between competing hypotheses. A further approach is to combine isotopes on multiple elements. For example, 15N, 18O and 17O can be measured on nitrate or nitrous oxide to understand the sources and cycling of these important N forms (Pérez et al., 2000; Costa et al., 2011), or parallel measurements on 13C and 15N in soil profiles can be used to trace the coupled cycling of C and N during soil profile development. For example, losses of 13C-depleted methane or 15N-depleted nitrous oxide as major fluxes will disproportionately increase the δ13C or δ15N values at specific soil depths favoring those two different processes. Further culture studies on ectomycorrhizal plants other than Pinus are needed, particularly of angiosperms, given the potential differences between angiosperms and gymnosperms in reliance on mycorrhizal fungi (Comas & Eissenstat, 2004). Culture studies on arbuscular mycorrhizal plants in which fractionation during mycorrhizal transfer of N from fungi to plants can be assessed are also desirable. Given the probable importance of organic forms of N to plant-available N (Inselsbacher & Näsholm, 2012), both culture studies and field studies should focus on exploring how organic N use in plants and mycorrhizal fungi influences isotopic patterns.

Acknowledgements

We thank Joseph Craine, John Hobbie, Andrew Ouimette, Colin Averill, and three anonymous reviewers for comments on previous versions. This work was supported by a grant from the US NSF Division of Environmental Biology to E.H., by a Bullard Fellowship to E.H. from Harvard University, and by grants from the Swedish Science Council to P.H. We thank Ronald Amundson, Joseph Craine, and Jordan Mayor for facilitating access to their respective databases.

Ancillary