Ectomycorrhizal fungal communities in two North American oak forests respond to nitrogen addition


Author for correspondence:
Peter G. Avis
Tel: +1 651 307 6327
Fax:+1 312 665 7158


  • • How nitrogen (N) deposition impacts ectomycorrhizal (EM) fungal communities has been little studied in deciduous forests or across spatial scales. Here, it was tested whether N addition decreases species richness and shifts species composition across spatial scales in temperate deciduous oak forests.
  • • Combined molecular (terminal restriction fragment length polymorphism (T-RFLP), sequencing) and morphological approaches were used to measure EM fungal operational taxon unit (OTU) richness, community structure and composition at the spatial scale of the root, soil core and forest during a 3-yr N fertilization experiment in Quercus-dominated forests near Chicago, IL, USA.
  • • In N treatments, significantly lower OTU richness at the largest but not smaller spatial scales and a different community structure were detected. The effects of N appeared to be immediate, not cumulative. Ordination indicated the composition of EM fungal communities was determined by forest site and N fertilization.
  • • The EM fungi responded to a N increase that was low compared with other fertilization studies, suggesting that moderate increases in N deposition can affect EM fungal communities at larger spatial scales in temperate deciduous ecosystems. While responses at large spatial scales indicate that environmental factors can drive changes in these communities, untangling the impacts of abiotic from biotic factors remain limited by detection issues.


Increased atmospheric nitrogen (N) deposition can greatly affect the biota (Stevens et al., 2004) and function (Magnani et al., 2007) of terrestrial ecosystems (Vitousek et al., 1997). For example, elevated N levels and/or their associated effects can alter mycorrhizal symbioses, the plant–fungal associations by which most plants acquire soil nutrients. Specifically, N increase can decrease ectomycorrhizal (EM) fungal species richness and shift the composition of EM fungal communities. Numerous studies (Karen & Nylund, 1997; Fransson et al., 2000; Jonsson et al., 2000; Lilleskov et al., 2001, 2002a; Peter et al., 2001; Edwards et al., 2004; Carfrae et al., 2006; Parrent et al., 2006) have reported the effects of N on EM fungi in conifer-dominated forest ecosystems. By contrast, only a few studies have examined the effects of increased N on EM in temperate deciduous ecosystems. Of these, two examined EM fungal communities along complex gradients of atmospheric pollutants including inorganic N (Baxter et al., 1999; Taylor et al., 2000) and one studied EM communities in an experimentally N-fertilized temperate oak savanna (Avis et al., 2003). As far as we know, no study has rigorously examined how N deposition affects the species richness and composition of EM fungal communities in temperate deciduous forests. This leaves a significant gap in our knowledge of the relationship of EM and N deposition as deciduous forests appear to differ from conifer forests in how they respond to increased N. Unlike conifer forests, N leaching in deciduous forests is difficult to predict and not closely linked to carbon (C) : N ratios or foliage N levels (Nilsson et al., 2007 and references therein). Considering that forests may depend on mycorrhizal fungi to retain N (Aber et al., 1998) and that many deciduous forests receive high levels of N deposition (e.g. ecosystems including forests in the Midwestern USA receive higher anthropogenic N than most other areas of the country; National Atmospheric Deposition Program, 2007), understanding how EM communities of deciduous forests respond to increased N is an exceptional concern.

The rates of atmospheric N deposition are expected to increase globally (Galloway et al., 2004; Phoenix et al., 2006), and this raises several questions. First, what level of N deposition will affect the species richness and composition of EM fungi in temperate deciduous forests? The few previous studies conducted in deciduous ecosystems do not provide an answer to this question. In oak forests in the eastern USA, Baxter et al. (1999) reported an approx. 40% decrease in EM fungal morphotype richness plus shifts in morphotype composition with about a tenfold increase in atmospheric nitrogen oxide concentrations. However, Taylor et al. (2000) observed no difference in morphotype diversity of beech forests in Europe across a similar gradient of N deposition. Although these studies make important contributions, they are limited because they lack the precision of species identifications afforded by molecular approaches for the documentation and interpretation of fungal diversity. Using a molecular-based approach Avis et al. (2003) measured a 15% drop in EM fungal richness belowground and a dramatic shift in the species composition in an oak savanna that was experimentally fertilized to an approx. 30-fold increase above ambient N for 15 yr. The level of N increases in the latter study and those in experimentally fertilized conifer forests were extreme and taken together provide little consensus for predicting how future realistic anthropogenic N increases will affect EM fungi or the roles EM serve in temperate deciduous forests in response to increased N.

Second, how quickly will EM fungi in temperate forests respond to elevated N deposition, and are the impacts on EM fungal communities immediate or gradual? In conifer-dominated ecosystems, changes can be quick and potentially accumulative. For example, in an experimentally N-fertilized spruce forest in Switzerland, Peter et al. (2001) found that diversity of EM sporocarps decreased dramatically after the first year of N fertilization but detected no change in belowground EM fungal diversity even after 2 yr of fertilization. In Alaskan spruce forests impacted for c. 30 yr by high levels of atmospheric N deposition from an industrial ammonia production facility, Lilleskov et al. (2001, 2002a) reported dramatic decreases in EM fungal richness above and below ground suggesting that changes will accumulate with continuous long-term N increase. However, no studies have examined these temporal effects of N in temperate deciduous forests, and the quality of response in these types of ecosystems remains unknown.

Third, at what spatial scales are changes in EM richness, community structure, and species composition detectable as result of N addition in temperate deciduous forests? Nitrogen deposition is expected to affect biota at a variety of spatial scales (Vitousek, 1994), but how spatial scale relates to the impact on EM communities by increased N deposition remains to be addressed. At smaller or micro-scales (e.g. mm2) such as root tips along a 1-cm long fragment of root, biotic interactions may drive EM fungal richness, structure, and composition (Bruns, 1995). For example, N supply can affect root growth dynamics (King et al., 2002) and fungal growth (Dickie et al., 1998; Lilleskov et al., 2002b), which may alter EM communities at small scales. At larger scales such as the forest stand (e.g. ha), however, environmental factors may dictate which species can exist as they do in plant communities (Reed et al., 1993; Fridley et al., 2006) and for other soil organisms (Ettema & Wardle, 2002). For example, increased N can decrease soil pH and alter nutrient availability from small to large scales (Fenn et al., 1998), Thus, interpretations of the mechanisms causing changes in EM fungal species richness and composition owing to N deposition will, at least in part, depend on understanding spatial scale and the ability to detect changes from the level of individual roots to the forest stand.

To address these questions, we tested the hypothesis that chronic, but relatively small increases in N (compared with other N addition studies), result in decreased EM fungal species richness and shifts in community structure and composition in temperate oak forests detectable at multiple scales. Since different fungal taxa likely vary in their responses to added N, some fungi may (or may appear to) tolerate and/or respond positively to additional N, while other taxa will be negatively affected. We base this hypothesis on the responses of EM fungi in other studies that tested much higher levels of experimental N increase in different woody plant communities (conifer and oak savanna).

We used an experimental fertilization approach that mimicked realistic future increases in N deposition. Our approach was unique, not only because of the relatively low concentration of N applied compared with other studies, but also because we applied N monthly in spatially and temporally specific amounts for 4 yr and as solutions of wet deposition rather than in a few highly concentrated dry applications. This application method addresses a problem of many fertilization studies (as pointed out by Britton & Fisher, 2007). In addition, we measured EM fungal species richness and composition at three spatial scales (the root, the soil core and the treatment plot) to examine if responses varied with respect to spatial scale. In this paper, we focus on the belowground species richness and composition of the EM fungal communities. Reports of aboveground responses (i.e. sporocarp data), specific responses of taxa belowground (e.g. percentage colonization by dominant EM colonizing roots) resulting in a change in community structure and ecosystem parameters (e.g. N cycling and plant responses) will be reported separately.

Materials and Methods

Study sites

The impact of N addition on EM fungi in two forests in the Chicago metropolitan area, Swallow Cliffs (SC) Forest Preserve in Cook County, IL, USA (41°40′29″N, 87°51′25″W) and Indiana Dunes (IND) National Lakeshore, Porter County, IN, USA (41°37′51″N, 87°05′13″W) were studied. Each site is a remnant of original forests (Braun, 1950), and tree composition is dominated by EM host plants, primarily white (Quercus alba) and red oak (Quercus rubra) growing on Morley siltloam, a fine illitic mesic Typic Hapludalf (Mapes, 1979; Furr, 1981). Basal area for both oak species is equal in both sites although SC has slightly higher stem density. National Atmospheric Deposition Program (NADP) analysis sites are within 2.5 km from each forest, and at the time the study started, average annual deposition of inline image and inline image at these sites was 19.27 and 20.87 kg ha−1 yr−1 at SC and IND, respectively (these amounts correspond to c. 7 kg N ha−1 yr−1 total). The amount of N deposition at these sites has remained relatively constant over the past 20 yr, including the years of this study.

Experimental fertilization

Four 0.1-ha plots (40 × 25 m) were established in May 2003 in each forest. Plots had similar EM host tree composition and were located at least 10 m from adjacent plots to minimize edge effect. Two of the four plots in each forest (i.e. SC and IND) represented controls and received no experimentally added N (i.e. only ambient deposition). The other two plots in each forest were spray-fertilized each month (12 applications per year) with three times the 5-yr average of nitrate and ammonium deposition measured at the NADP site near each respective forest for that month (the total amount of inline image and inline image added to fertilized plots since fertilization started in 2003 is approx. 180 kg ha−1). Fertilizer solutions consisted of the site and monthly appropriate concentrations of potassium nitrate and ammonium sulfate (BFG Supply, Burton, OH, USA) dissolved in 6 l of deionized water for each 0.1 ha treatment plot. Control plots did not receive a water control as the 6 l of solution added to the experimental plots during application was negligible.

Nested sampling of EM fungi belowground

The EM fungi colonizing the roots of oaks were sampled from each plot annually in the first or second week of August 2004–06. In each 0.1 ha plot, 10 mature oak (Q. alba or Q. rubra) trees (> 30 cm diameter at breast height (dbh) and at least 3 m apart; estimated to be at least 90-yr old; Jones et al., 2006) were randomly designated as ‘focal trees’. One soil core (10 cm diameter and 14.5 cm deep) was collected 1.5 m from the base of each focal tree each year. Cores were taken from the same ‘focal tree’ in successive years. In years 2 and 3 cores were taken at least 1 m from the previous year's core but still 1.5 m from the focal tree. Soil cores were placed in a cooler and then stored at 4°C at The Field Museum until analyzed further.

Sampling of EM fungi colonizing roots generally followed Avis et al. (2003) but included a nested survey design to allow for comparisons of EM response to N across different spatial scales. Briefly, all roots from each core were removed by hand from unwashed cores. These roots were then gently washed with water to remove remaining soil and debris, cut into 2–3 cm lengths in a Pyrex dish containing water, and mixed thoroughly. From the pool of roots, five root fragments were selected randomly and the first 10 root tips encountered on each fragment were determined to be mycorrhizal or not. These 50 root tips represented the smallest spatial scale (0.000039 m2; on average, each tip is c.  1 mm long and 1 mm diameter) at which EM fungal species richness and composition were measured. If mycorrhizal, the EM tip was placed into a morphological category (i.e. ‘morphotyped’ using a dissecting microscope) based on EM mantle features such as color, texture and surface ornamentation (Agerer, 1987–96; Goodman et al., 1996). Representative samples of each morphotype were preserved for both DNA analysis (see later) and as vouchers on microscope slides. After the first 50 tips were counted, the remainder of the roots in each core were scanned under a dissecting microscope to see if additional morphotypes could be found. Any additional morphotypes encountered at the scale of the entire soil core (0.008 m2; 10 cm diameter, 14.5 cm deep) survey were also preserved for molecular identification and as a voucher. Data from the 10 cores of each plot (0.08 m2 collected in the 1000 m2; 0.1 ha plots) were then combined to examine patterns at the largest spatial scale that could be replicated.

To identify each morphotype on a molecular basis, the internal transcribed spacer (ITS) region of ribosomal DNA was analysed by a combination of terminal restriction fragment length polymorphism (T-RFLP) and sequencing. A single tip of each morphotype collected per core was dried either by freezing in liquid N and then drying in a SpeedVac (similar to Avis et al., 2003) or by placement of the EM tip near silica gel (similar to Koide et al., 2005). DNA from each of the dried tips was then extracted following Avis et al. (2003) using a plant DNA extraction kit (REDExtract-N-Amp Plant polymerase chain reaction (PCR) Kit, Sigma, St Louis, MO, USA) except that three times the suggested dilution solution was used. The ITS region was amplified using the primer combination of ITS1-F (Gardes & Bruns, 1996) labeled with 6FAM from Integrated DNA Technologies, Inc. (Coralville, IA, USA) and ITS4 (White et al., 1990) labeled with VIC from Applied Biosystems (Foster City, CA, USA). Thermocycling conditions were: 94°C for 85 s (1 cycle); 95°C for 35 s, 55°C for 55 s, 72°C for 45 s (14 cycles); 95°C for 35 s, 55°C for 55 s, 72°C for 2 min (15 cycles); 95°C for 35 s, 55°C for 55 s, 72°C for 3 min (10 cycles); 72°C for 10 min. The PCR products were visualized and then excised and purified by Gelase (Epicentre Biotechnologies, Madison, WI, USA) or purified separately with QIAquick PCR Purification Kit (Qiagen, Inc., Mississauga, Ontario, Canada). A T-RFLP analysis followed Avis et al., (2006) and utilized restriction digestions by HinfI and HaeIII (New England Biolabs, Inc., Beverly, MA, USA). Fragment patterns were determined using an ABI 3730 DNA analyzer and genemapper software (Applied Biosystems). Analysis of T-RFLP results used TRAMPR (Fitzjohn & Dickie, 2007) implemented in R version 2.4.0. To identify the fungi colonizing roots in each core, we modified the ‘database’ T-RFLP approach described by Dickie & Fitzjohn (2007). TRAMPR was used to match the T-RFLP pattern of individual root tips to a database of previously determined T-RFLP patterns from sporocarps and EM roots (Avis et al., 2006) or from ITS sequences (sequencing and cloning if necessary followed; Avis et al., 2006) from EM fungi colonizing roots collected in this study.

Molecular and morphological data were combined to assign identity to the EM fungi sampled and we refer to these identifications as operational taxonomic units (OTUs) rather than species, T-RFLP types or morphotypes. This molecular–morphological ‘fusion’ approach was necessary because not all EM tips sampled produced reliable molecular identifications because of a lack of ITS amplification or detectable T-RFLP (e.g. owing to low concentration of amplified DNA). This approach was conservative and provided a baseline from which to interpret effects of N deposition on EM fungi in our study sites.

We used a set of rules to assign EM fungi to an OTU. (1) If a single EM root tip produced only one T-RFLP, the tip was considered to be a unique OTU as long as the T-RFLP did not match that of another morphotype in the core. (2) If a T-RFLP of a morphotype matched that of a different morphotype in the core, then the original morphotype designation was considered inappropriate. Frequency (and root tip colonization) for that sample was then merged for the two or more morphotypes sharing the T-RFLP. (3) If more than one T-RFLP was produced by a single EM tip, each unambiguous pattern was counted. However, if the T-RFLP was inconsistent with the sample's morphotype (e.g. a Cenococcum morphotype produced a T-RFLP equal to that from Russula amoenolens), the T-RFLP was considered to be from a nontarget fungus (i.e. other EM tips in the core or other fungal tissue such as spores or hyphae adhered to the tip). In this case, the additional T-RFLP was not assigned to that tip. (4) If no T-RFLP was produced by an EM root tip sample, then identification was based only on morphotype. If the morphotype was unambiguous and well known (e.g. > 20 tips morphotyped as ‘Yellow 2’ were found to produce a T-RFLP from Boletus rubellus), a matching tip was considered to be from that respective OTU even without T-RFLP confirmation. If the morphology was not interpretable (e.g. the sample was called ‘Russuloid 1’ but other morphotypes in the core included ‘Russuloid 2’) then we invoked parsimony and considered both morphotypes equal and they were placed in the ‘Russuloid’ OTU. This latter situation undoubtedly grouped numerous species together, but allowed us to conservatively estimate the richness of each sample. These rules were rigorously employed for samples from all 10 cores per plot each year.

Statistical analysis

The effects of N fertilization, site and year, and their respective two and three way interactions, on EM fungal richness measured at the three spatial scales (based on presence or absence of an OTU) were tested by two-way anova with Tukey–Kramer HSD for multiple comparisons and assumptions of normality using jmp software version 4.0 (SAS Institute, Inc., Cary, NC, USA).

The rates at which species were found in our sampling were determined by producing OTU-sampling unit (i.e. roots, soil cores, plot) curves to compare ambient and fertilized plots from the two forest sites. Curves were generated using pcord, version 4 (McCune & Mefford, 1999) with subsampling repeated 500 times per soil core. Ninety-five percent confident intervals were computed to determine if N fertilization affected the rates at which OTU were collected (Hughes et al., 2001).

The EM fungal community composition was compared in N-fertilized and unfertilized plots and between the two forest sites by using nonmetric multidimensional scaling (NMS) implemented in pcord (McCune & Mefford, 1999). Taxa were relativized within each plot by the most abundantly encountered EM fungal OTU in each plot (i.e. using the ‘relativization by maximum’ function in pcord). This standardization decreases the influence of dominant species (P. Minchin, pers. Comm.). Default settings were used for NMS (Sorenson's/Bray–Curtis distance measure, 400 maximum iterations, 0.00001 for the instability criterion, 6 starting axes, 40 real runs and 50 randomized runs). In addition, community composition was related to the amount of N deposited on each plot by canonical correspondence analysis (CCA) using pcord. For the environmental variables used to constrain the ordination of species composition data, ‘sites’ were coded as a quantitative ‘dummy’ variable and the log transformed amount of ammonium and nitrate deposited onto ambient (i.e. control) or N-fertilized plots was determined for each year (2004–06). Row and column scores were standardized by Hill's method. Monte Carlo permutation tests using 999 permutations were used to test the strength of relationship between species composition and the variables of site and N deposition.


Terminal-RFLP and ITS sequence types

The EM fungi on oak roots from 240 soil cores (40 per site from two sites in each of 3 yr) colonizing c. 400 000 root tips were surveyed (c. 1500–2000 EM tips per core at these sites; percent colonization by EM fungi averaged 90% and did not differ between control and N-fertilized plots or sites; total number of ectomycorrhizal root tips, root length and biomass did not vary among plots; see the Supplementary Material, Table S3). In total, 12 000 root tips were morphotyped and molecular analyses was conducted on 1333 individual EM root tips. Of these tips, 94.4% (1249) were successfully PCR amplified. Of these, 884 produced T-RFLPs that were interpreted; and 365 PCR amplifications either did not produce complete T-RFLPs or were too ambiguous to interpret. Eighty-three EM tips produced more than one set of T-RFLP patterns. While some of these might represent tips containing multiple fungi, the high sensitivity of the T-RFLP method can also detect additional T-RFLPs (e.g. from nontarget fungal tissue adhered to an EM tip or from intra-collection ITS variation; Avis et al., 2006).

A total of 314 T-RFLP types; with 153 in two or more cores; and 58 in four or more cores were recorded during the 3 yr of the study (Table 1 and the Supplementary Material, Table S1). Forty T-RFLPs from EM root samples collected below ground matched T-RFLPs of sporocarps collected above ground (in a nonexhaustive database of c. 100 species; see Avis et al., 2006) and 19 additional T-RFLPs were identified with ITS sequences and blast searches. A total of 266 T-RFLPs, mostly singletons, were designated as ‘unknown’ OTUs. While some of these unidentified T-RFLPs might come from nontarget fungi (see previous paragraph), this number is likely low. All tips were initially screened via a morphotyping step, thereby ensuring that all root tips analysed had an EM mantle and most of the fungal DNA extracted would probably have come from mantle tissue. In addition, blast searches returned EM fungi as the top matches for all of the T-RFLP types sequenced (Table 1).

Table 1.  Ectomycorrhizal fungi (EM) found in more than three soil cores over 3 yr (2004–06)
Ectomycorrhizal fungiIND AmbIND +NSC AmbSC +NMorphotypeAccessionTop blast match% Similarity
  1. Data in columns 2–5 are number of cores in which that EM was found. The table is ordered into four groups based on which site–treatment combination an EM was found most frequently. Column headings: ‘Ectomycorrhizal fungi’, the particular identification of the EM. If matched by terminal restriction fragment length polymorphism (T-RFLP) to a sporocarp, then genus and species of sporocarp are provided along with collection number. If not matched to a sporocarp but identified by TRFLP and if also by internal transcribed spacer (ITS) sequence, TRFLP identifier is given as well as any taxonomic identification based on blast search and phylogenetic analysis. ‘IND’, ‘Indiana Dunes’; ‘SC’, ‘Swallow Cliffs’. ‘Amb’, control plots. ‘+N’, fertilized plots. ‘Morphotype’, names refer to those in the Supplementary Material Table S2, each defined by color, mantle surface type and other distinguishing features, respectively. Many morphotype names encompass species complexes which are represented by multiple operational taxon units (OTUs) in our data set (e.g. Cenococcum and Lactarius quietus). If ITS sequences are included, the GENBANK accession number is provided as well as top blast match and similarity score.

Boletus rubellus PRL598014 712 1Yellow 2   
TRFLP 16 Cenococcum10 8 310Cenococcum   
Russula mariae group 1 PRL6228 9 4 1 7Russuloid   
Russula brevipes PRL5981 8 0 0 0Russuloid w/c   
TRFLP 647 Cenococcum 7 0 1 0CenococcumEU375704Uncultured EM (DQ474368)369/375 (98%)
TRFLP 196 Thelephoraceae 5 2 2 2TomentelloidEU375705Uncultured EM (DQ377421)468/469 (99%)
TRFLP 57 5 0 5 3Sebacinoid   
TRFLP 708 Clavulinaceae 5 0 0 0Russuloid   
TRFLP 12 Tomentella sp. 4 4 0 0TomentelloidEU375707Uncultured EM (AY874387)470/486 (96%)
TRFLP 120 3 4 0 0Cortinarioid   
TRFLP 17 3 3 3 3RussuloidEU375708Cantharellaceae (EF619643)562/563 (99%)
TRFLP 27 Pezizaceae 3 2 0 RussuloidEU427527Peziza (EF411091)572/585 (97%)
Russula pectinatoides PRL5406 3 1 0 1Russuloid w/c   
TRFLP 272 Inocybe sp. 3 0 2 0TomentelloidEU427526Inocybe (AM882994)386/391 (98%)
TRFLP 559 Sebacina sp. 3 0 2 2SebacinoidEU375709Sebacina sp. (UDB000773)524/550 (95%)
TRFLP 668 Clavulinaceae 3 0 2 0Russuloid   
TRFLP 67 3 0 1 0Russuloid   
TRFLP 158 Russula pectinatoides group 2 1 2 2Russuloid w/cEU375710Basidiomycete (AY970218)542/546 (99%)
Entoloma sp. PRL5252 2 1 1 1Cortinarioid   
Gyroporus castaneus PRL5872 2 1 1 1Orange 1   
Russula silvicola PRL5877 2 0 2 0Russuloid   
TRFLP 167 2 0 2 0Russuloid   
Inocybe albodisca PRL5343 1 1 1 1Russuloid   
TRFLP 31ac 1 1 1 2Cortinarioid   
IND AmbIND +NSC AmbSC +N    
Russula amoenolens PRL5798 611 4 7Russuloid w/c   
TRFLP 2 Cenococcum 512 3 2Cenococcum   
TRFLP 4 Russula mariae group 2 6 9 2 8RussuloidEU375711Russula sp. (AF418629)614/633 (96%)
TRFLP 31b 5 9 3 4Cortinarioid   
TRFLP 14 Inocybe sp. 2 6 1 0RussuloidEU375712Uncultured EM (AJ893275)293/317 (92%)
TRFLP 712 2 5 0 2Tuberoid   
TRFLP 26 Tomentella sp. 1 4 2 3TomentelloidEU375713Uncultured EM (EF411072)626/639 (97%)
TRFLP 125 0 4 2 0Tuberoid   
TRFLP 153 Russula sp. 0 4 0 0Russuloid w/cEU375714Russula tenuiceps (DQ974756)627/658 (95%)
Entoloma sp. PRL5282 and 6180 0 4 0 0Russuloid   
Russula sp PRL5248 2 3 1 3Russuloid   
TRFLP 159 2 3 0 0Tomentelloid   
TRFLP 151 0 3 0 0Sebacinoid   
TRFLP 33 0 1 5 4Sebacinoid   
TRFLP 577 0 1 2 2Russuloid   
IND AmbIND +NSC AmbSC +N    
Russula sp. PRL3666c group 0 310 6Russuloid   
TRFLP 11 3 3 7 4Tomentelloid   
TRFLP 327 Lactarius quietus group 0 0 7 7Russuloid Lage   
TRFLP 106 Tomentella sp. 1 4 6 2TomentelloidEU375715Tomentella sp. (DQ835998)392/400 (98%)
Scleroderma areolatum PRL5883 2 3 5 3Sclerodermoid   
TRFLP 320 Lactarius quietus group 2 3 5 2Russuloid Lage   
Inocybe rimosa PRL5296 0 0 3 2Tuberoid   
TRFLP 797 0 0 3 2Russuloid   
IND AmbIND +NSC AmbSC +N    
TRFLP 242 Lactarius quietus group 0 2 812Russuloid Lage   
TRFLP 15 Pezizomycotina 6 6 611RussuloidEU375716Pezizomycotina (DQ273333)389/403 (96%)
Cortinarius Telamonia group PRL5107 1 1 210Cortinarioid   
TRFLP 214 0 0 1 5Tomentelloid   
TRFLP 199 Hebeloma sp. 2 2 2 4CortinarioidEU375717Uncultured EM (AY730685)612/612 (100%)
Entoloma sp. PRL5673 1 1 0 4Cortinarioid   
Russula amoenolens PRL6219 0 0 0 4Russuloid w/c   
TRFLP 404 Lactarius hygrophoroides 2 1 2 3Yellow 1EU375718Lactarius longisporus (DQ421971)463/504 (91%)
TRFLP 228 1 0 2 3Tuberoid   
TRFLP 219 0 0 2 3Tomentelloid   
TRFLP 882 0 0 2 3Russuloid   

Effect of N fertilization and year on EM fungal richness across spatial scales

The effects of N fertilization and year on EM fungal OTU richness appeared to differ across the spatial scale investigated. An effect of N fertilization on richness was detected at the scale of the 1000 m2 treatment plot, (Table 2), with significantly fewer OTUs in the fertilized plots than in the control plots at both sites. The effect of year was also significant with fewer OTUs found in each of the successive years of the study on average regardless of treatment. However, OTU richness did not accumulatively decrease with added N in each year. At the root scale, where a mean of 3.6 EM fungal OTU was detected, richness did not differ significantly between ambient and N treatments, sites or years. At the scale of the soil core, richness differed significantly by year (F1,1 = 22.5, P = 0.0002) with 6.5, 5.6, and 3.9 OTUs in 2004, 2005, and 2006, respectively, but no N treatment or site differences were detected. Interaction effects were not significant at any of the spatial scales measured.

Table 2.  Mean ectomycorrhizal (EM) fungal richness in control (ambient) and nitrogen (N)-fertilized (+N) treatment plots at two oak forest sites over 3 yr, and the corresponding anova results of the effects of fertilization, site and year
YrMeanTreatment plot scale (1000 m2)
Swallow CliffsIndiana Dunes
  • *

    All df = 1,1; ns, not significant.

anova source
  F*P value
N fertilizer 13.10.0023
Site nsns
Year 10.20.0057

The difference in fungal richness between treatments resulted from detection of fewer infrequently found OTUs in the fertilized plots. The total number of singletons at a site differed significantly (t-test, P = 0.03) between treatments with a combined average of 90 singletons in ambient plots and 61.5 singletons in fertilized plots. The number of singletons did not differ between sites (t-test, P = 0.80). Doubletons did not differ significantly between treatments or sites.

The rate at which we detected new OTUs was decreased by N fertilization, as indicated by OTU-sampling unit curves (Fig. 1). Curves for ambient plots from both forest sites (SC and ID) were significantly steeper and did not overlap (95% confidence intervals) with those from fertilized plots when data from soil cores from all 3 yr were combined (Fig. 1). Similar curves were produced when curves were generated for each individual year (not shown). The OTU sampling unit curves integrate across the spatial area from which samples are collected and thus provide an indicator of what spatial scale N influences EM species richness. In Fig. 1, treatment versus control plot curves begin to diverge above five soil cores suggesting that the scale of impact is about half a treatment plot or c. 500 m2.

Figure 1.

The 2004–06 operational taxonomic unit (OTU)-sampling unit (soil core) curve. Curves are mean ectomycorrhizal (EM) fungal OTU richness for soil cores determined by resampling the EM fungal richness up to 500 times in soil cores collected over 3 yr in ambient and nitrogen (N) fertilized Indiana Dunes (ID) and Swallow Cliffs (SC) plots. Error bars represent standard deviations for the up to 500 resampling iterations.

The influence of site and N fertilization on EM fungal OTU composition

At the treatment plot scale, NMS ordination clustered the OTU composition by site along axis 1 and by N fertilization along axis 2 (not shown). These axes significantly reduced the stress of the ordination (axes 1 and 2, each P = 0.02) and explained 47% of the variation in the data. This was consistent with the CCA (Fig. 2), which showed that site and N fertilization were strongly related to the composition of EM fungal communities. The total variance of the OTU composition summarized by treatment plots was 6.0311 with the eigenvalue of 0.519 for axis 1 and 0.318 for axis 2, which represented 8.6% and 5.3% of the total variation, respectively. Monte Carlo permutation tests indicated that axis 1 (P = 0.006, r = 0.983) was significantly related to environmental variables. Ordinations using data summarized at the spatial scale of soil core (i.e. 240 cores) did not reveal any clustering by site or treatment (not shown).

Figure 2.

The 2004–06 ectomycorrhizal (EM) fungal community canonical correspondence analysis (CCA) ordination of treatment plots in two experimentally nitrogen (N)-fertilized oak forests. Circles and the letter ‘I’ represent plots in Indiana Dunes; diamonds and the letter ‘S’ represent Swallow Cliffs. The first number of each plot label indicates the year surveyed. The last number indicates plot number. Symbols with the letter ‘N’ or ‘C’ next to them are from N-fertilized or control plots, respectively. The length of the overlay vectors from the center of the diagram indicates the increasing strength of relationship between a variable and the species composition of the plots. Overlain are log nitrate and log ammonium added to these plots.

The frequency of EM fungi in Table 1 complements the ordinations and further describes the influence of site and N fertilization on EM communities. Some OTU were only found in certain sites. For example, T-RFLP 12 Tomentella sp. and TRFLP 327 were found numerous times but only at the SC site. In addition, EM fungi appeared to respond differentially to fertilization. TRFLP 272 Inocybe and TRFLP 668 Clavulinaceae, although found in both sites, were only found in ambient plots. The most frequently encountered OTU, Boletus rubellus PRL5980, was more frequently encountered in control plots while Russula amoenolens PRL5798 was more often encountered in fertilized plots.

Temporal effects

The treatment effect on EM fungal richness was immediate as a significant difference was detected after only 1 yr of fertilization (Table 2). The composition of EM fungal communities also appeared to shift immediately since differences between treatments were found in 2004.

The effect of year on EM fungal richness was significant at the two larger spatial scales sampled. This effect appeared to result from substantial yearly variation in richness rather than from a cumulative decrease in richness with fertilization. Annual variation in OTU composition was also high within plots and within sites.


Our study supports the hypothesis that EM fungal communities are less species-rich and differ in structure and composition when N deposition is elevated in temperate oak forests. This effect was found even though the amount of added N was low compared with other studies (see the Introduction) suggesting that, at least in temperate oak forests, EM fungi are sensitive to changes in N deposition within the range of predicted future anthropogenic deposition. In our study, the EM fungal communities were less OTU-rich when fertilized despite community composition differences between sites (Fig. 2). The decrease in richness was attributable to decreases in singleton (i.e. infrequently encountered or ‘rare’) fungi in fertilized plots at both sites.

We observed an approx. 20% drop in fungal richness with only a three fold increase in experimental N deposition in these two forests. This is similar to the responses by EM fungal richness in other temperate deciduous ecosystems that received greater proportional increases in N deposition along pollution gradients (Baxter et al., 1999) or much higher N addition treatments (Avis et al., 2003). In comparison to studies conducted in conifer forests exposed to increased N, our results are similar to those of Jonsson et al. (2000) where treatment plot EM fungal species richness was c. 20% lower in plots receiving threefold greater experimental N deposition. The responses in our study were not as dramatic as those of Lilleskov et al. (2002a) who reported about a 70% decrease in EM species richness with a tenfold increase in N deposition. Although comparisons between conifer and deciduous forests may be limited, but taken together these studies suggest that even moderate increases in N deposition can impact the EM fungal richness, at least at larger spatial scales in temperate deciduous forests.

The response by EM below ground to experimental N deposition at our sites appears to be immediate but not cumulative. The effect of N fertilization was pronounced after only 1 yr of treatment but no treatment by year interaction effect was significant suggesting that there was no increasing effect of N over the time of our study. One explanation for this result is that a set of relatively rare fungi was unable to colonize roots under these levels of elevated N. This hypothesis fits our observations but is contrary to the data provided in the only other study that examined initial responses of EM fungal richness to elevated N (Peter et al., 2001); this reported little change in the EM colonizing roots in the first 3 yr of an experimentally fertilized spruce forest. If the results from our study hold for deciduous forest ecosystems as a whole, an immediate loss of N-sensitive EM fungi would be one way that deciduous forests and those dominated by conifers respond to elevated N.

Although we attempted to simulate N deposition very closely, our study design was limited in at least two ways. First, we only examined the impact of an immediate increase in N deposition rather than a ramped increase over time. This could be a concern since studies that have compared ramped vs instantaneous increases in environmental impacts on mycorrhizas such as those comparing the effects of ramped vs step increases in CO2 on arbuscular mycorrhizal fungi (Klironomos et al., 2005) have shown dramatic differences in effects. However, unlike the increases in global CO2, increased N deposition may be more step-wise considering the relatively rapid shifts in local and regional industrial and agricultural practices that lead to N deposition (e.g. a steep but relatively local N deposition gradient adjacent to a fertilizer factory; Lilleskov et al., 2001). As a result, our N application may be more similar to these kinds of anthropogenic step increases in N. Second, we applied the N fertilizer using a backpack sprayer and sprayed the soil, any plants and debris from the height of c. 1 m to the ground. The forest canopy was therefore not affected and the responses we see here are not attributable to any indirect effects of foliar N uptake by canopy trees. Interestingly, a recent study that used a helicopter to test how N applied to a forest canopy influenced N cycling (Gaige et al., 2007) identified large differences between canopy treatments vs ambient inputs and suggested that N uptake by the canopy foliage was important. However, since the helicopter study was conducted in a conifer-dominated forest it remains to be seen how deciduous forests would respond. Foliar uptake would likely account for a significant proportion of N uptake by deciduous trees when in leaf. Given the lack of foliage during winter in our study forests, our fertilization design might closely mimic what occurs for half the year. In any case, it is likely that foliar N uptake would exacerbate the effect documented. Thus our data are conservative estimates of the impact of increased anthropogenic N deposition.

Primarily relying on T-RFLPs rather than sequence data, for identifying EM fungal OTUs likely reduced the resolution of our data. As in all methods that rely on DNA fragment patterns, distinct T-RFLPs could be from different individuals of the same species resulting in an overestimation of richness (Karen et al., 1997; Horton, 2002; Avis et al., 2006; Smith et al., 2007) while different species might display similar T-RFLPs causing richness to be under-sampled (Horton & Bruns, 2001). Although 266 OTUs in our data set remain unknown, the lack of non-EM fungi identified among the subset of OTUs sequenced and the matches of T-RFLP-morphotype based OTUs to T-RFLP patterns of EM fungal sporocarps from our sites suggests that the number of spurious T-RFLPs from nontarget fungi in our data set is low. A small number of spurious OTUs would not change the general patterns documented in this study.

Effects attributable to factors other than N

Since we used fertilizer solutions that consisted of potassium nitrate and ammonium sulfate, we cannot be certain that the responses we observed are caused only by N. Both potassium and sulfate could have an effect on EM fungi as potassium and sulfur are important nutrients and sulfate can also be a pollutant. However, studies examining the impact of sulfate on EM communities in spruce forests showed it had no effect on EM fungal species richness (Karen & Nylund, 1996; Carfrae et al., 2006). We are not aware of any studies that examine the impact of potassium on EM communities.

Changes in root tip density have been proposed to co-occur with changes in fungal diversity in some studies that consider the impact of N fertilization on EM communities (Taylor, 2002). However, we did not observe a shift in root density as a result of increased N in a parallel study conducted in these same fertilization plots (see the Supplementary Material, Table S3) and, therefore, have no evidence that a shift in root density is influencing the observed change in EM fungal richness. This result is consistent with other recent studies where root density was found to be unrelated to N fertilization in oak woodlands. Nadelhoffer (2000), Avis (2003) and Phillips & Fahey (2007) all show that fertilization did not decrease the density of fine roots in oak-dominated temperate ecosystems. For each of these studies, the amount of N added was far greater than that added in our study lending further support to our contention that changes in EM fungal richness are caused by factors other than shifts in root tip density.

Responses to N at larger spatial scales

We detected an effect of N on richness, structure and composition of EM fungal communities at larger spatial scales (> 500 m2) only. While Baxter et al. (1999) and Avis et al. (2003) noted differences caused by increased N at larger spatial scales, our study is the first to systematically address the relationship of spatial scale and the impact of N on EM fungi. However, we cannot rule out the possibility that N had an effect at small scales that went undetected – detection of fungi at small spatial scales requires addressing ‘sampling’ issues wherein fewer organisms are contained in smaller areas, which leads to increased variance via sampling error (Reed et al., 1993; Taylor, 2002). Advances in ecological analysis using hierarchical linear models (McMahon & Diez, 2007) and maximum likelihood methods (Fridley et al., 2006) will benefit future investigations that address scale and the impact of N deposition on EM fungi.


We report that EM fungal richness was lower and species structure and composition shifted with moderate increases of experimental N deposition in temperate deciduous oak forests. These changes were detectable at scales > 500 m2, suggesting that environmental effects can drive these differences. Changes occurred during the first year of N addition, but EM fungal richness and EM fungal species composition between treatments did not diverge further over successive years of treatment indicating that the effects were not cumulative during the 4 yr of this study. Given the projected future increases in N deposition, it will be important to include these potential responses into the predictions of how EM fungi in temperate deciduous forest ecosystems will respond.


We thank P. Leacock, H. BassiriRad, H. Sehtiya, M. Jorgensen, W. Gaswick, P. Hogan, K. Dean, C. Keene, G. Williams, P. Watson, G. Kaufman and K. Lauer, The Field Museum and the Pritzker Laboratory for Molecular Evolution and Systematics for their assistance. Valuable comments were provided by I. Alexander and anonymous reviewers. Funding was provided to G.M.M. and J.L. by NSF DEB 0134748.