Fungal functioning in a pine forest: evidence from a 15N-labeled global change experiment



  • We used natural and tracer nitrogen (N) isotopes in a Pinus taeda free air CO2 enrichment (FACE) experiment to investigate functioning of ectomycorrhizal and saprotrophic fungi in N cycling.
  • Fungal sporocarps were sampled in 2004 (natural abundance and 15N tracer) and 2010 (tracer) and δ15N patterns were compared against litter and soil pools.
  • Ectomycorrhizal fungi with hydrophobic ectomycorrhizas (e.g. Cortinarius and Tricholoma) acquired N from the Oea horizon or deeper. Taxa with hydrophilic ectomycorrhizas acquired N from the Oi horizon (Russula and Lactarius) or deeper (Laccaria, Inocybe, and Amanita). 15N enrichment patterns for Cortinarius and Amanita in 2010 did not correspond to any measured bulk pool, suggesting that a persistent pool of active organic N supplied these two taxa. Saprotrophic fungi could be separated into those colonizing pine cones (Baeospora), wood, litter (Oi), and soil (Ramariopsis), with δ15N of taxa reflecting substrate differences. 15N enrichment between sources and sporocarps varied across taxa and contributed to δ15N patterns.
  • Natural abundance and 15N tracers proved useful for tracking N from different depths into fungal taxa, generally corresponded to literature estimates of fungal activity within soil profiles, and provided new insights into interpreting natural abundance δ15N patterns.


In most forests, fungi are key players in the below-ground cycling of nitrogen (N) and carbon (C), with saprotrophic fungi mainly decomposing surface litter and woody debris and ectomycorrhizal fungi more active in organic and mineral horizons below the litter layer (Lindahl et al., 2007). Yet information on how ectomycorrhizal fungi differ in exploration for N is scarce. In one approach that may provide some insight into how ectomycorrhizal taxa differ in their N acquisition strategies, Agerer (2001) proposed that ectomycorrhizal fungi could be classified based on the extent and method of hyphal exploration of the soil, with fungi possessing hydrophilic ectomycorrhizas usually classified into contact, short-distance, or medium-distance smooth exploration types, and fungi with hydrophobic ectomycorrhizas generally classified into medium-distance fringe and long-distance exploration types. Peay et al. (2011) proposed that these exploration types may correlate with strategies for acquiring C from roots, with medium- and long-distance exploration types more likely to colonize roots at greater depth or further from trees than short-distance exploration types. Exploration types may be adapted for specific N forms, with taxa with hydrophobic ectomycorrhizas focused on insoluble forms of N such as protein or chitin, and taxa with hydrophilic ectomycorrhizas focused on soluble forms of N (Trudell et al., 2004; Hobbie & Agerer, 2010; Lilleskov et al., 2011).

Nitrogen isotopes (15N : 14N, expressed as δ15N) have been a useful tool in exploring fungal functioning. For example, several studies have demonstrated that saprotrophic fungi are lower in δ15N than ectomycorrhizal fungi, presumably for two reasons: saprotrophic fungi assimilate primarily wood-derived or litter-derived N (15N-depleted) and ectomycorrhizal fungi assimilate deeper soil N (15N-enriched) (Kohzu et al., 1999; Hobbie, 2005); ectomycorrhizal fungi transfer 15N-depleted N to their host plants, leading to 15N-enriched fungal biomass (Hobbie & Colpaert, 2003). Hobbie & Agerer (2010) observed that δ15N patterns in sporocarps correlated with hydrophobicity of ectomycorrhizas, with hydrophobic exploration types c. 3‰ higher in 15N than hydrophilic exploration types. Soil δ15N increases with depth (Billings & Richter, 2006; Hobbie & Ouimette, 2009), and Agerer et al. (2012) correlated hyphal exploration depth in Ramaria taxa with fungal δ15N. The higher δ15N in hydrophobic than in hydrophilic ectomycorrhizal fungi may accordingly indicate that hydrophobic taxa are active at deeper depths than hydrophilic taxa.

One way to assess whether source N or transfer of 15N-depleted N controls fungal δ15N is to use 15N labeling to generate data that can be compared against natural abundance patterns. 15N labeling should change the δ15N of N sources but not change 15N effects of internal partitioning and N transfer to host plants. Therefore, by comparing natural abundance and tracer results, differences in the source δ15N of different fungal taxa can be estimated, because we can assume that the 15N enrichment (ε) between sporocarps and source N is the same in both cases. This information can then be used to estimate the 15N enrichment between sources and sporocarps and the probable sources for sporocarp N. Although tracer 15N labeling studies have also been useful to study N dynamics in many terrestrial ecosystems (Currie et al., 1996; Hofmockel et al., 2011; Wang & Macko, 2011), 15N labeling has yet to be examined systematically in sporocarps of different fungal taxa, despite the potential insights into both fungal functioning and the interpretation of natural abundance δ15N.

Here, we used information on natural abundance and tracer 15N patterns in sporocarps and other ecosystem pools at the Duke free air CO2 enrichment (FACE) study in a Pinus taeda forest. Although we report differences in some cases between elevated and ambient CO2 treatments in δ15N of different pools, we do not focus on treatment effects of elevated CO2. Instead, we use the 15N-applied label to provide general insights into fungal functioning that are unavailable from studies solely at natural abundance.

At the Duke FACE site, the forest floor was labeled with tracer concentrations of 15N in 2003. Ecosystem samples were collected in 2003, 2004, 2005, and 2010, allowing the 15N label to be tracked over time as it was assimilated by biota and migrated from surface to deeper horizons. In soil collected from 2003 to 2005, 15N labeling patterns were initially high in surface layers and then decreased over time, while increasing over time at greater depths (Hofmockel et al., 2011). We suggest that fungal 15N labeling patterns will reflect two factors: the δ15N of the soil and litter pools from which fungi obtain N; and 15N enrichment between sources and sporocarps arising from 15N and 14N partitioning within fungi. Therefore, measuring fungal δ15N at two different time points under changing background concentrations of 15N labeling allows the second factor to be accounted for and can improve estimates of the soil horizons from which fungi acquire N (Fig. 1). This can be expressed mathematically in the following equations, where δ15Nft1 and δ15Nft2 are the fungal signatures at time t1 and t2, δ15Nst1 and δ15Nst2 are the source signatures at time t1 and t2, and ε is the 15N enrichment between sources and sporocarps.

Figure 1.

Schematic of 15N movement through the soil profile and into fungi at: (a) natural abundance, (b) 1 yr after 15N labeling, and (c) 7 yr after 15N labeling. Relative 15N for soil horizons and fungi within each diagram indicated by colors, as follows: light blue, low 15N; tan, medium 15N; brown, high 15N.

display math(Eqn 1)

Multiplying both sides by (1 + δ15Nst1) and solving for δ15Nft1 we get:

display math(Eqn 2)

A similar equation can be written for the fungal signature at time t2:

display math(Eqn 3)

Subtracting Eqn 3 from Eqn 2 gives:

display math(Eqn 4)

By measuring isotopic patterns in ecosystem pools and fungi at the Duke FACE site, we tested the following hypotheses:

  1. Saprotrophic fungi will generally use N sources found close to the surface and ectomycorrhizal fungi generally use deeper N. Ectomycorrhizal fungi with hydrophobic ectomycorrhizas will use N from deeper depths than taxa with hydrophilic ectomycorrhizas.
  2. Both source differences and 15N enrichment relative to sources contribute to 15N differences among different fungal taxa, with 15N enrichment relative to sources greater for ectomycorrhizal fungi than for saprotrophic fungi.

Materials and Methods

FACE experiment

The FACE experiment at the Duke Forest (35°58′41″N, 79°05′39″W 163 m elevation, Orange County, NC, USA) comprised six 30-m-diameter plots. Three experimental plots were fumigated with CO2 to maintain the atmospheric CO2 concentration at 200 ppm above ambient (i.e. 565 ppm at the start of the experiment). Three control plots were fumigated with ambient air only (365 ppm at the start of the experiment). The experiment began on 27 August 1996, and was continuous during daylight hours until fumigation ended on 1 November 2010. Additional details on the FACE operation are available in Hendrey et al. (1999) and at

The Duke Forest originated from 3-yr-old loblolly pine (P. taeda) seedlings that were planted in 1983 in a 2.4 × 2.0 m spacing. In 1996, the 16-yr-old pine trees were c. 14 m tall, had a density of 1600 stems ha−1, and comprised 98% of the basal area of the stand. A deciduous understory layer consisted primarily of sweetgum (Liquidambar styraciflua), with some red maple (Acer rubrum), redbud (Cercis canadensis), and dogwood (Cornus florida). Topographic relief is < 1° throughout the 32 ha site. Soils are derived from mafic bedrock and classified as Enon Series (fine, mixed, active, thermic Ultic Hapludalfs). Soils are slightly acidic (pH = 5.75 in 0.1 M CaCl2), and have well-developed soil horizons with mixed clay mineralogy. Mean annual temperature is 15.5°C and mean annual precipitation is 1140 mm. Additional site details are available in Schlesinger & Lichter (2001) and Finzi et al. (2001).

15N labeling and fertilization experiment

In May 2003, 98 atom% 15N was applied via backpack sprayers in each of the six plots, with 75% of the 15N applied as NH4Cl and 25% as KNO3 at a rate of 15 mg 15N m−2 in 0.25 l H2O. This represented 3% of the inorganic N pool (0–15 cm depth).

Starting in 2005, each of the six FACE plots was divided in two, and one-half of each was fertilized with N in a split-plot design. Here, we only used samples from the unfertilized plot sections.

Sporocarp collection

Sporocarps were collected in 2004 on 14 October, 25 October, and 27 October from ambient plots (15N tracer), elevated CO2 plots (15N tracer), and from outside the experimental plots (ambient CO2, natural abundance 15N). Sites were thoroughly surveyed and all sporocarps collected and identified. In 2010 sporocarps were collected from ambient and elevated CO2 plots from 30 October until 3 December. The 2010 sampling was not quantitative. Fungal sporocarps were either air-dried or flash-frozen in the field and freeze-dried. Taxa were identified from macroscopic and microscopic morphological characteristics. Taxa were further classified as to whether they possessed hydrophobic or hydrophilic ectomycorrhizas (Agerer, 2006; Di Marino et al., 2008). We have assumed that hydrophobicity of ectomycorrhizas is a characteristic that is conserved at the genus level (Unestam & Sun, 1995; Agerer, 2006).

Soil and other ecosystem pool collection and analysis

The Oi (forest litter), Oea, 0–15 cm, and 15–30 cm soil horizons were sampled in March 2003 (natural abundance) and in September 2003–2005 (15N tracer; Hofmockel et al., 2011). In addition to bulk analyses, dissolved organic N (DON), microbial biomass, and fine roots (< 2 mm) were analyzed for %N and δ15N. Identical sampling protocols were used in September 2010 for bulk analyses.

Samples from 2003 to 2005 were analyzed according to Hofmockel et al. (2011), with isotopic analyses at the UC Davis Stable Isotope Facility. Analyses of 2010 samples were treated similarly, but were analyzed at the UNH Stable Isotope Laboratory. Natural abundance δ15N was measured on whole sporocarps from outside the FACE plots collected in 2004. For tracer 15N measurements in FACE rings, we used whole sporocarps in 2004 and the average of sporocarp caps and stipes in 2010. Values are reported in the standard notation (δ15N; ‰) relative to atmospheric N2 for N, where δ15N = (Rsample/Rstandard − 1) and R is the molar ratio of 15N : 14N (Lajtha & Michener, 1994).

Statistical analysis

JMP (SAS, Cary, NC, USA) was used for statistical analysis. We used linear regression models to explore the factors regulating isotopic composition of fungal sporocarps. Data were not log-transformed, as fungal N isotopes passed a normality test for 2004 and 2010 samples (Shapiro–Wilk goodness of fit, = 0.071 and 0.292, respectively). Regression models in JMP explaining δ15N values in 2004 were run separately for natural abundance and tracer data sets and for saprotrophic and ectomycorrhizal fungi. Linear regression models for δ15N values in 2010 combined saprotrophic and ectomycorrhizal sporocarps, with CO2 treatment and genus as the explanatory variables. Because %N can correlate with sporocarp δ15N at natural abundance (Hobbie et al., 2012), potential explanatory variables at natural abundance included genus, hydrophobicity, CO2 treatment, and N concentration, whereas tracer 15N data sets did not include N concentration. For the categorical variables of CO2 treatment and hydrophobicity, JMP assigned a value of −1 for elevated CO2 and hydrophobic taxa, with ambient CO2 and hydrophilic taxa receiving a value of +1. Interaction terms were included in models unless they resulted in singular terms. To test for statistical differences between two pools, t-tests assuming unequal variance were used.

Linear regression models of δ15N in soil profiles included horizon and CO2 treatment as explanatory variables. For statistical comparisons of δ15N in 2004 among fine roots, DON, and microbial biomass in different horizons, paired t-tests were used to compare specific soil pools, with CO2 treatment included as an explanatory variable.


Natural abundance isotopic patterns in soil and fungi

Isotopic values for bulk soil at natural abundance as sampled in March 2003 under ambient and elevated CO2 are presented in Table 1. In a regression model that included horizon and CO2 treatment, horizon affected δ15N (< 0.001), with the Oi horizon the lowest and the mineral soil at 15–30 cm the highest. CO2 treatment did not affect soil δ15N (= 0.284). Complete statistical results are given in Supporting Information, Table S1.

Table 1. δ15N (in ‰) in ambient and elevated CO2 plots at the Duke Forest free air CO2 enrichment (FACE) site for litter and soil pools, 2003–2010 (± SE)
YearOiOea0–15 cm15–30 cm
  1. a

    Natural abundance values, sampled in March 2003, before 15N labeling.

  2. b

    September 2003, after 15N labeling.

2003a−5.0 ± 0.3−4.2 ± 0.2−3.0 ± 0.3−3.1 ± 0.42.6 ± 0.32.0 ± 0.86.6 ± 0.45.5 ± 0.2
2003b340.3 ± 31.5362.5 ± 32.090.9 ± 6.887.6 ± 27.87.1 ± 0.36.8 ± 0.614.0 ± 0.515.6 ± 1.8
200451.9 ± 8.847.2 ± 14.474.4 ± 14.267.6 ± 9.75.6 ± 1.14.7 ± 0.512.4 ± 0.312.2 ± 1.3
200514.6 ± 1.512.5 ± 1.267.9 ± 9.159.4 ± 3.611.7 ± 1.88.2 ± 1.04.8 ± 0.66.6 ± 0.4
201012.2 ± 2.210.3 ± 1.318.5 ± 2.418.0 ± 0.812.3 ± 1.411.7 ± 1.09.8 ± 0.38.2 ± 0.3

Fungi were sampled for natural abundance in 2004 outside the FACE plots. At natural abundance δ15N, ectomycorrhizal taxa were higher than saprotrophic fungi (t-test, < 0.001; 2.9 ± 0.3‰, = 94 vs −0.5 ± 0.2‰, = 70). Among ectomycorrhizal taxa, Amanita and Russula were the lowest and Sistotrema the highest. Taxa with hydrophobic ectomycorrhizas were significantly higher in 15N than were taxa with hydrophilic ectomycorrhizas (t-test, < 0.001; 6.9 ± 0.3‰, = 17 vs 2.1 ± 0.2‰, = 77). In saprotrophic fungi, sporocarp δ15N varied with substrate, with the soil fungus Ramariopsis the highest, litter decay fungi intermediate, and the cone-colonizing fungus Baeospora and wood decay fungi the lowest (Fig. 2). In a regression model of sporocarp δ15N that included sporocarp %N and genus, %N significantly affected δ15N of saprotrophic fungi but not ectomycorrhizal fungi, whereas genus significantly affected δ15N in both ectomycorrhizal and saprotrophic fungi (Table 2). In saprotrophic fungi, %N accounted for 16% and genus accounted for 84% of the explained variance. In Fig. 2, %N and δ15N are plotted for different genera of saprotrophic and ectomycorrhizal fungi, showing that δ15N is generally in the order hydrophobic ectomycorrhizal taxa > hydrophilic ectomycorrhizal taxa > saprotrophic taxa, with %N generally higher in saprotrophic taxa than in ectomycorrhizal taxa.

Table 2. Regression model of δ15N of saprotrophic and ectomycorrhizal fungi as a function of genus and sporocarp %N at natural abundance in 2004
ParameterValue ± SE P ParameterValue ± SE P
  1. Only data on sporocarps from outside free air CO2 enrichment (FACE) rings were used in the model. Adjusted r2 was 0.529 (= 70, < 0.001) for saprotrophic fungi and 0.574 (= 94, < 0.001) for ectomycorrhizal fungi for the entire model. Values were estimated from the regression models analyzed for δ15N. Preferred substrates for saprotrophic fungi are indicated in parentheses after the genus name, as follows: C, pine cones; L, leaf litter; W, wood; S, soil. Hydrophobicity of ectomycorrhizas is indicated after the genus, as follows: phi, hydrophilic; pho, hydrophobic. Table S3 has complete statistical information. Statistical significance (< 0.05) is indicated in bold.

Intercept−2.5 ± 0.80.004Intercept4.7 ± 1.10.004
%N0.42 ± 0.18 0.022 %N−0.12 ± 0.270.660
Genus 0.003 Genus < 0.001
Genus-specific effects
Baeospora (C)−0.9 ± 0.50.098Amanita (phi)−2.8 ± 0.5 < 0.001
Clitocybe (L)−0.9 ± 0.70.181Cortinarius (pho)2.9 ± 0.7 < 0.001
Gymnopilus (W)−0.7 ± 0.40.067Hebeloma (pho)0.8 ± 1.80.602
Marasmius (L)−1.1 ± 1.20.365Inocybe (phi)−1.5 ± 0.8 0.031
Mycena (L)0.5 ± 0.50.322Laccaria (phi)−1.0 ± 0.90.191
Pholiota (W)−0.6 ± 0.50.241Lactarius (phi)−0.3 ± 0.70.618
Pluteus (W)0.0 ± 0.60.976Russula (phi)−2.5 ± 0.4 < 0.001
Ramariopsis (S)2.5 ± 0.8 0.002 Sistotrema (pho)3.3 ± 0.8 < 0.001
Rhodocollybia (L)1.1 ± 0.5 0.033    
Figure 2.

Natural abundance δ15N plotted against %N (± SE) of different genera of ectomycorrhizal and saprotrophic fungi collected outside of free air CO2 enrichment (FACE) rings across all plots. Data are from 2004 collections as reported for %N in Table S2. Closed blue symbols, ectomycorrhizal with hydrophobic ectomycorrhizas; closed red symbols, ectomycorrhizal with hydrophilic ectomycorrhizas; open symbols, saprotrophic. Values for Pinus litter, Pinus cones, and wood are also plotted (± SE).

Tracing 15N labeling in soil and fungi

After tracer application in May 2003, δ15N values in the Oi horizon were initially very high (c. 350‰), but declined dramatically over the next 2 yr (Table 1). By contrast, the high values in the Oea horizon in 2003 (c. 90‰) declined more slowly. The low δ15N values of the shallow mineral soil horizon (0–15 cm) increased slightly from 2003 to 2005, while decreasing over this period in the 15–30 cm horizon. δ15N values in fine roots, DON, and microbial biomass in 2004 were much higher in the Oea horizon than in the 0–15 cm and 15–30 cm horizons (Fig. 3). In the Oea horizon, the δ15N order was roots << DON < microbes, with all three pools differing significantly from one another in paired t-tests (Table S3). These three pools did not differ significantly in paired t-tests at 0–15 cm, but at 15–30 cm microbes were again higher than roots (paired t-test, = 6, = 0.001). None of these pools differed by CO2 treatment.

Figure 3.

Nitrogen (N) isotope values for fine roots, dissolved organic N, and microbial biomass in 2004 under ambient (amb) and elevated (elev) CO2 treatments after 15N labeling (± SE). The sampling horizon is indicated in the figure by closed circles (Oea), open circles (0–15 cm), or closed triangles (15–30 cm). Statistically significant differences in paired t-test comparisons within a horizon are shown by lowercase letters for Oea and uppercase letters for 15–30 cm, with different letters indicating differences.

By 2010, soil δ15N was lower in the Oi horizon than in the Oea horizon, and then further decreased in the mineral soil (Table 1). Depth (< 0.001), but not CO2 treatment (= 0.358), affected soil δ15N (Table S1).

Application of 15N tracer in 2003 greatly affected the δ15N of sporocarps collected in 2004. Both symbiotic status (= 0.002) and CO2 treatment (= 0.009) significantly affected sporocarp δ15N, with no significant interaction between these two variables (= 0.084). Saprotrophic fungi were significantly higher in δ15N in elevated CO2 treatments than in ambient treatments (t-test, = 0.008), but ectomycorrhizal fungi were not (t-test, = 0.191) (Fig. 4). In ambient CO2 plots, ectomycorrhizal fungi averaged 69.9 ± 7.2‰ (= 33) and saprotrophic fungi averaged 79.8 ± 8.3‰ (= 36), whereas in elevated CO2 plots, ectomycorrhizal fungi averaged 76.5 ± 6.1‰ (= 45) and saprotrophic fungi averaged 112.1 ± 7.9‰ (= 32). The δ15N among saprotrophic genera did not differ significantly, whereas the ectomycorrhizal genera Lactarius and Russula were significantly higher than several other ectomycorrhizal genera, including Inocybe, Laccaria, and Cortinarius (complete statistical model given in Table S4). Sporocarps with hydrophilic ectomycorrhizas (e.g. Lactarius, Russula) averaged 20.8‰ higher in δ15N than sporocarps with hydrophobic ectomycorrhizas (e.g. Cortinarius) (t-test, = 0.034).

Figure 4.

Nitrogen (N) isotopes of different genera of ectomycorrhizal and saprotrophic fungi (± SE) in ambient and elevated CO2 treatments in 2004. A 1 : 1 line (dashed) is also shown. Closed blue symbols, ectomycorrhizal with hydrophobic ectomycorrhizas; closed red symbols, ectomycorrhizal with hydrophilic ectomycorrhizas; open symbols, saprotrophic. Rhodo., Rhodocollybia; Gymno., Gymnopilus.

In 2010, 30 sporocarps were collected from unfertilized plots. In a model exploring the factors influencing fungal δ15N in 2010, genus affected fungal δ15N (= 0.041) but CO2 treatment did not (= 0.577). Some taxa with hydrophilic ectomycorrhizas were high in δ15N (28‰ for Amanita and 27‰ for Laccaria) and some were low (15‰ for Russula and 18‰ for Lactarius), whereas both taxa with hydrophobic ectomycorrhizas were high in δ15N (39‰ for Cortinarius and 24‰ for Tricholoma). Hydrophobic taxa averaged 10.7‰ higher than hydrophilic taxa (t-test, = 0.004). A single Hygrophorus was at 17.9‰ and a single Rhodocollybia was at 15.8‰. Between 2004 and 2010 in 15N-labeled plots, Lactarius declined by 72‰, Russula by 60‰, and the saprotrophic Rhodocollybia by 56‰. Cortinarius declined 15‰ on average (Figs 4, 5).

Figure 5.

Comparison of δ15N values (± SE) of five taxa of fungi and four soil pools between natural abundance samples in 2003 (soils) and 2004 (fungi)and 15N-enriched samples in 2010 (soils and fungi). Lines of constant 15N enrichment between 2004 and 2010 samples are shown as dotted lines between 5 and 30‰.

When δ15N values of soil pools and of sporocarps at natural abundance in 2004 and in 2010 are plotted, we can estimate the δ15N of the source N if we assume that 15N enrichment from source to sporocarp is the same for 2004 and 2010 samples (Eqns 1-4). In the resulting Fig. 5, lines of constant 15N enrichment between 2004 and 2010 samples were used to suggest potential sources. The patterns indicate that, between 2004 and 2010, 15N enrichment of Rhodocollybia, Lactarius, and Russula was similar to the Oi horizon and 15N enrichment for Laccaria was similar to the Oea horizon, whereas 15N enrichment values for Amanita and Cortinarius were greater than that of any measured bulk pool.


Depth distribution of fungal activity

Because supplies of energy-rich C are largest in the surface litter, saprotrophic fungi are commonly found there and should obtain N from those sources (Lindahl et al., 2007). By contrast, ectomycorrhizal fungi acquire labile C from plant sugars, and therefore have a relative advantage over free-living heterotrophs in nutrient acquisition as the energy content of organic matter declines at deeper depths. Ectomycorrhizal fungi in the surface litter may rely on enzymatic breakdown by free-living saprotrophic microbes (Lindahl et al., 2005), whereas those active at greater depths presumably rely on their own enzymatic capabilities. Within ectomycorrhizal fungi, the enzymatic capabilities to acquire N from polymeric substances such as protein and chitin appear greater for hydrophobic than for hydrophilic taxa (Lilleskov et al., 2011). Accordingly, taxa with hydrophobic ectomycorrhizas should acquire N from greater depth and, because soil δ15N increases with depth, be higher in δ15N than taxa with hydrophilic ectomycorrhizas.

To assess whether different fungal groups acquired soil N from different depths, we compared natural abundance δ15N in fungal taxa in 2004 with soil in 2003 and compared fungi with soil from 15N-labeled plots sampled in 2003–2005 and in 2010 (Table 1, Fig. 5). In apparent support of our first hypothesis, fungal δ15N patterns at natural abundance were at their lowest in saprotrophic fungi, intermediate in hydrophilic ectomycorrhizal fungi, and at their highest in hydrophobic ectomycorrhizal fungi, presumably reflecting the use of N from shallow litter horizons or wood by saprotrophic fungi and two taxa of hydrophilic ectomycorrhizal fungi (Lactarius and Russula), and assimilation of deeper soil N by both hydrophilic (Laccaria and Amanita) and hydrophobic ectomycorrhizal taxa.

After plots were labeled with 15N in 2003, δ15N values were initially very high in the leaf litter (c. 350‰), but quickly declined (c. 50‰ by 2004) as new litter of low δ15N was added, 15N-labeled litter decomposed, and its N was incorporated into the Oea horizon (Table 1). In 2004, the year following application of the tracer, fungal δ15N followed expected patterns, with δ15N of saprotrophic fungi > hydrophilic ectomycorrhizal fungi > hydrophobic ectomycorrhizal fungi, again reflecting use of shallow soil N for saprotrophic and hydrophilic ectomycorrhizal fungi and use of deeper soil N for hydrophobic (and some hydrophilic) ectomycorrhizal fungi. However, Russula, Lactarius, and most of the saprotrophic fungi appeared to at least partially sample the 2003 litter cohort, as their δ15N values in 2004 in 15N-labeled plots were higher than any measured bulk pool for that year but lower than Oi values for 2003. Alternatively, DON and microbial N may have supplied N to these fungi, as they generally had similar δ15N values as the DON and microbial N in the Oea horizon in 2004 (Fig. 3); unfortunately, DON and microbial biomass were not measured in the Oi horizon, so precise comparisons are not possible. In contrast to these patterns, the hydrophilic taxa Inocybe, Amanita, and Laccaria had lower δ15N values than DON or microbial N from the Oea horizon, suggesting a possible contribution from mineral horizon N (Fig. 4, Table 1).

Finally, in 2010, 7 yr after 15N tracer was applied, soil and fungal δ15N patterns had changed, with the applied 15N migrating deeper down the soil profile. Between 2004 and 2010, fungal δ15N decreased more in the saprotrophic Rhodocollybia and in the hydrophilic Lactarius and Russula than in the hydrophobic Cortinarius and the hydrophilic Amanita and Laccaria. This indicates that these latter two taxa assimilated deeper N than Lactarius and Russula but did not increase in 15N relative to sources as much as other ectomycorrhizal taxa. These patterns were consistent with saprotrophic fungi and some hydrophilic ectomycorrhizal fungi using shallow soil N. In addition, the hydrophobic Tricholoma in 2010 was also high in δ15N, suggesting that taxa with hydrophobic ectomycorrhizas generally explore deeper horizons. High δ15N in Tricholoma have been reported before and attributed to high 15N enrichment between the mycelia and sporocarps (Zeller et al., 2007); here, our results indicate that acquisition of deeper, 15N-enriched N partially explains the high δ15N values of Tricholoma sporocarps.

To infer potential N sources, we paired δ15N measurements in 2004 at natural abundance with 2010 measurements from the FACE rings (Fig. 5). We assumed comparable 15N enrichment between sources and fungi for both sampling times. Lines of constant 15N enrichment for 2010 relative to 2004 indicate the probable δ15N at natural abundance of the source N, which could then be compared against the measured bulk soil pools. Based on high 15N labeling patterns in 2010 and the assumed similar 15N enrichment relative to sources in 2004 (natural abundance) and 2010 (15N tracer applications), Amanita, Cortinarius, and Laccaria acquired N from deeper in the soil profile (where 15N is higher) than Lactarius, Russula, and the litter-inhabiting Rhodocollybia. According to Eqn 4, the δ15N differences from 2004 to 2010 between sources and between sporocarps are not exactly equivalent, as they are related by the factor (1 + ε). However, if the estimated 15N enrichment value (ε) is 20‰, then they would only differ by 0.2‰ in 15N enrichment.

High 15N enrichments (between 25 and 32‰) of 15N-labeled 2010 values relative to natural abundance values in 2004 for Amanita and Cortinarius (as indicated by isolines on Fig. 5) are higher than any estimated soil pool enrichment. This implies that these two taxa access N pools that have retained more 15N than any measured bulk soil pool. Although additional pools other than bulk were not measured in 2010, the δ15N values in 2004 in the FACE plots for DON and microbial N were c. 30‰ higher than concurrently measured bulk values, indicating that pools such as these could potentially be similarly elevated relative to bulk pools in 2010. Such differences in δ15N between bulk pools and other soil fractions have previously been reported from 15N labeling experiments (Zeller & Dambrine, 2011), and presumably reflect variability in turnover times of different biochemical or biophysical fractions.

Resource partitioning in saprotrophic fungi

The saprotrophic taxa colonized different substrates, with Baeospora myosura commonly colonizing conifer cones (Rayner et al., 1985), Gymnopilus, Pholiota, Pluteus, and Oligoporus commonly colonizing wood (Pouska et al., 2010), and Mycena, Rhodocollybia, Marasmius, and Clitocybe commonly colonizing litter (Rayner et al., 1985). Baeospora and the wood decay fungi were similar in natural abundance δ15N (Fig. 2), suggesting that these taxa are using N of similar δ15N, with that N presumably relatively unprocessed and drawn from undecayed pine cones or wood. However, we point out that the litter decay fungi Clitocybe and Marasmius had similar values to Baeospora and the wood decay fungi in the regression model that included %N (Table 2). Some of the δ15N variability among taxa should accordingly be driven by variable proportions of 15N-enriched protein and 15N-depleted chitin, with higher %N and δ15N presumably correlating with higher relative protein content (Hobbie et al., 2012).

Pine cones closely matched Baeospora in 2004 (natural abundance), with a measured 15N enrichment in Baeospora relative to pine cones of 1.4‰. Higher δ15N values for the litter decay fungi Rhodocollybia and Mycena than for Baeospora, wood decay fungi, and the other two litter decay taxa suggest some degree of 15N enrichment during initial stages of decay of litter and subsequent acquisition of this N by Rhodocollybia and Mycena. The relatively high δ15N in Ramariopsis may indicate foraging for N in deeper soil horizons than other saprotrophic taxa. Whereas the other saprotrophic taxa feed on wood or fresh litter, Ramariopsis kunzei is typically reported as fruiting on ground, often buried below the surface litter (Arora, 1986), implying that it obtains N from deeper sources than fresh litter.

Depth distribution of ectomycorrhizal fungi and comparisons with other studies

Studies that have examined the depth distribution through soil profiles of ectomycorrhizal fungi can provide additional insights. In general agreement with the patterns reported here, some studies detected taxa of Lactarius and Russula in litter horizons or coarse woody debris, with Amanita, Cortinarius, Inocybe, and Laccaria found in deeper horizons (Dickie et al., 2002; Landeweert et al., 2003; Tedersoo et al., 2003). Although patterns were not as clear in other studies (summarized in Table S5), we note that fungal presence is unlikely to correlate precisely with fungal N source. For example, the density of the N resource will generally be greater in organic horizons than in mineral horizons, and therefore N uptake per unit of fungal biomass may be greater in organic horizons.

Some indication of the greater spatial extent of exploratory hyphae in taxa with hydrophobic vs hydrophilic ectomycorrhizas can be inferred from the study of Genney et al. (2006), in which fungal identity of both colonized root tips and of extraradical hyphae were assessed. In this study, 88% of the Cortinarius observations were of extraradical hyphae, compared with only 31% of the Lactarius rufus observations and 34% of the Cenococcum geophilum observations. Cortinarius has hydrophobic ectomycorrhizas, whereas Lactarius and Cenococcum have hydrophilic ectomycorrhizas. Similarly, a reanalysis of data in Peay et al. (2011) from Pinus-associated communities at forest edges and interior locations (10 m inside the edge) for the genera found at Duke FACE indicates that 77% of hydrophilic ectomycorrhizas were found at interior locations, compared with only 46% of hydrophobic ectomycorrhizas, with interior locations presumably associated with less extensive extraradical hyphal development than edge locations. As suggested in Hobbie & Agerer (2010), the apparently greater spatial exploration in hydrophobic than in hydrophilic taxa may lead to higher relative sequestration of 15N-depleted chitin in transport structures (extraradical hyphae) and the formation of 15N-enriched N within fungi that is metabolically available for sporocarp formation, therefore contributing to 15N enrichment of hydrophobic relative to hydrophilic taxa.


We conclude that saprotrophic fungi primarily assimilate N found in the Oi horizon (with the possible exception of Ramariopsis), and fungi with hydrophobic ectomycorrhizas primarily assimilate N from deeper horizons such as the Oea, supporting Hypothesis 1. However, taxa with hydrophilic ectomycorrhizas appeared to assimilate from the Oi horizon in some cases (e.g. Russula and Lactarius) and in other cases from deeper horizons (e.g. Amanita and Laccaria), which only partially supports Hypothesis 1. These differences in N acquisition largely corresponded to N isotope patterns at natural abundance.

Estimated 15N enrichment relative to bulk substrates was in the order: Lactarius Russula > Rhodocollybia > Baeospora, and Cortinarius Amanita. It appears that 15N enrichment relative to source is variable among taxa, and therefore contributes, along with source δ15N, to 15N differences among taxa in sporocarps, which supports Hypothesis 2. Higher δ15N values in some fungi than in any measured ecosystem pool suggest that fungi access N pools that are not captured in the relatively gross characterizations of litter and soil δ15N distributions presented here. In previous work, 15N enrichments have been analytically linked to partitioning of N between plants and fungi, within different fungal structures, such as extraradical hyphae vs sporocarps, or between protein and chitin (Hobbie et al., 2005, 2012). The results here suggest one approach to generate the data needed to test these analytical approaches.


This work was supported by grant DEB-1146328 from the US National Science Foundation (NSF), by a Bullard Fellowship to E.H. from Harvard University, and by grant ER65430 from the Department of Energy to K.H. Core funding for the Duke FACE site was provided by the Office of Science (BER), US Department of Energy, grant no. DE-FG02-95ER62083. NSF also supported the work on 15N tracer application and redistribution (DEB-0236356 and DEB-0235425). We thank Matthew Henn for assistance with collection of sporocarp samples. We thank Amy Sojka for graphics assistance and thank Lixin Wang, Christian Rixen, and six anonymous reviewers for comments on previous versions of this manuscript.